KneeDaily

1.1

Bundle Structure of the Anterior Cruciate Ligament

Yongtao Maov1.0Published 29 references
Evidence through 2026.10.01Chinese original
Contents

The anterior cruciate ligament (ACL) is composed of large numbers of collagen fibers that run continuously from the femoral to the tibial insertion. Under the microscope, collagen fibrils assemble into fiber subunits about 20 μm in diameter, which in turn form fascicles tens to hundreds of micrometers across; between fascicles lie membranous tissues that are discontinuous and of uneven thickness. The "bundles" referred to in gross anatomy and surgery are larger fiber groups made up of several fascicles. A membranous septum that runs the full length of the ligament and is constantly present could divide it into two or three structures, each with its own bounding membrane; such a septum has not been demonstrated in the adult ACL.

The double-bundle, triple-bundle and flat ribbon descriptions each rest on reproducible observations: anterior and posterior fibers tighten alternately with flexion and extension; in some specimens the anteromedial portion can be further divided into medial and lateral parts; the ligament is flat in cross section in extension and twists as a whole in flexion. Differences among studies in the number of bundles reported are related to several factors: the angle and viewing direction of the knee at observation, whether and how the synovium is removed, where blunt separation begins, who makes the judgment, and which reconstruction method the study was intended to support anatomically. The effects of the separation path (Zhao and Huangfu, 2010) and of the arthroscopic viewing portal (Steckel et al., 2009) have each been compared directly in the same set of specimens or the same set of patients; the independent contribution of each factor has not been adequately compared.

In the author's view, the anteromedial (AM), posterolateral (PL) and intermediate (IM) bundles are suitable as names for fiber regions that mark the relative positions of fibers on the femoral and tibial insertions and their tension behavior with flexion and extension; the available anatomical evidence does not support treating them as structures that can be counted one by one and that are independently enveloped.

When this section speaks of a "bundle" without qualification, it means a partition at the level of AM, PL and IM; groupings defined by individual studies keep the names used in the original. Precise localization of the insertions, blood supply and innervation are covered in Sections 1.2–1.4, and the quantitative contributions of each fiber region to anterior and rotational stability in Sections 1.5 and 1.6.

1. Hierarchical Organization of the Fibers

Clark and Sidles (1990) first fixed the knee in situ with glutaraldehyde, then removed the ligament and examined it by light microscopy and scanning electron microscopy, avoiding the shrinkage and deformation that follow excision. Human specimens came from 6 donors aged 64–67 years; canine and rabbit specimens were also studied. In all three species, collagen fibrils assembled into fiber subunits about 20 μm in diameter, separated by rows of cells; several subunits formed fascicles 20–400 μm or more in diameter, enclosed by membranous septa within which vessels ran exclusively (pp. 184, 187). These membranes were discontinuous and of uneven thickness, and the authors could not rigorously identify or track larger bundle groups on that basis; larger groupings appeared to exist in the human ACL, but no continuous bounding membrane separated them (p. 182, legend to Fig. 2C). Blunt dissection could not completely separate the fascicles; mesentery-like tissue folds still connected them to the main body of the ligament. Viewed from the front, the ACL resembled two bands of similar size; from another angle this impression disappeared, and the cross section was heart-shaped because of a depression on the anterior surface (p. 181). The authors concluded that the ACL has a tendency to form two "ill-defined bands," anteromedial and posterolateral, and that this tendency is a phenomenon of fascicle grouping (p. 187).

Odensten and Gillquist (1985) studied 33 normal cadaveric knees; donors had died at 17–52 years of age (mean 28.7 years), all from accidents, and the number of donors was not reported. In 20 knees examined grossly, after removal of all non-collagenous tissue, no anatomical subdivision was seen (p. 258); in another 3 knees serial whole-joint sections were cut, and 5 excised ACLs were sectioned transversely; under picrosirius red staining and polarized light, no bundling was seen either, and between the collagen fibers lay evenly distributed, vascularized connective tissue (p. 260, Fig. 6 at ×14). The same study recorded that at 90° of flexion the ligament twisted about 90°. The authors did not on that account deny functional subdivision; the Discussion states that fiber length and tension change with flexion, a result consistent with the view that the ACL has functionally distinct parts (p. 261).

Fuss (1989) studied 22 knees in all (16 fixed, 6 fresh); in 20 of them the cruciate ligaments were artificially divided into small bundles and their origins and insertions traced separately. No pre-existing septum was seen during separation—neither many fine septa nor two or three coarse ones (pp. 165–166, 169). He classified fibers by whether the distance between origin and insertion changed during flexion and extension into thin "guiding bundles" of constant tension and "limiting bundles" that tighten near terminal extension, a functional grouping. Mommersteeg et al. (1995) digitized 3 ACLs in three dimensions (donors aged 63–89 years), first measuring the course of superficial fascicles and then bluntly separating fascicles of the same direction into bundle groups; each ACL required 8 or 9 bundle groups to preserve the principal fiber directions (p. 463). The three ligaments were partitioned differently, and the authors state that the grouping was not based on fibrous tissue septa (p. 467). The often-cited "6–10 bundles" includes the posterior cruciate ligament; the ACL itself accounts for 8–9, the minimum number of groups needed to preserve the principal fiber directions, not a count of independent structures. In the same study a cleft was seen posteriorly in all 3 ACLs, corresponding to the posterior boundary between the anteromedial and posterolateral portions, whereas the anterior boundary was unclear (p. 463); adjacent bundle groups were not necessarily adjacent over the full course from femur to tibia (p. 464).

Parrilli et al. (2024) used micro-CT at 17.5 μm voxel size to display both ACLs of a single 60-year-old donor: in extension, multiple fascicles arranged in parallel; in flexion, fascicles twisted upon one another. This was a methodological demonstration with only 1 donor and cannot be used to estimate bundle number.

Key points

  • Two microscopic levels: Collagen fibrils assemble into fiber subunits about 20 μm in diameter, which in turn form fascicles 20–400 μm across; the membranous septa between fascicles are discontinuous, and larger groupings in the human ACL have no continuous bounding membrane (Clark and Sidles, 1990).
  • Groupings above the fascicle: Gross dissection and polarized-light sections (Odensten and Gillquist, 1985), artificial separation (Fuss, 1989) and three-dimensional grouping (Mommersteeg et al., 1995) all failed to find a pre-existing septum running the full length of the ligament. Mommersteeg's 8–9 bundle groups are the minimum needed to preserve fiber direction, and the three ligaments were partitioned differently.
  • Functional subdivision does not presuppose anatomical septa: Studies that found no bundling likewise recorded that fiber length and tension change with flexion and extension.

2. Origins and Reliable Elements of the Double-Bundle Description

The double-bundle description is usually traced to Girgis et al. (1975). That study dissected 20 cadaveric knees and 24 fresh knees, and the original description differs markedly from the two-bundle descriptions in later work. The authors wrote that the ACL is flat in extension and that in flexion the ligament winds and twists; fibers arising from a flange of the femoral insertion form a thin single band that tightens in flexion and is the shortest part of the ACL (p. 221). At 120°–130° of flexion the femoral insertion turns from vertical to horizontal, the main body of the ligament becomes lax, and only the fibers arising from the flange and forming the thin anteromedial band remain tight (pp. 223–224). The summary of the paper characterizes the ACL as a distinct anteromedial band and a main posterolateral portion (p. 230). The original double bundle was a thin band and the remaining bulk, distinguished by tension and markedly unequal in size; the description of a flat ligament in extension also preceded the later "ribbon" concept by nearly forty years.

Later studies have mostly described the two parts as two bundles of similar size that can be measured separately. Steckel et al. (2007) measured midsubstance cross-sectional areas of 20.3 mm² for the AM bundle and 17.7 mm² for the PL bundle, with mean lengths of 37.7 mm for AM and 20.7 mm for PL (pp. 389–390, Tables 1 and 2); this length relationship is the opposite of Girgis's description of the anteromedial band as the shortest part. The two descriptions are difficult to map onto the same set of fibers. Before comparing bundle lengths, areas and tension data across studies, one must first establish how each study made its divisions.

What stands up to repeated testing in the double-bundle description is the distribution of tension. Girgis et al. observed by hand, Amis and Dawkins (1991) measured length changes of fiber portions with flexion and extension using fine wires in 9 knees, and Steckel et al. (2007) observed on a testing machine; the direction of the results is consistent: fibers of the anteromedial portion tighten in flexion, and fibers of the posterolateral portion tighten in extension. Fuss (1989) and Odensten and Gillquist (1985), who saw no anatomical bundling, also described this functional difference. The specific contributions of each portion are covered in Sections 1.5 and 1.6.

The review by Petersen and Zantop (2007) is often cited as anatomical support for the double bundle. The degree of certainty in the Abstract and the body of the text is inconsistent: the first sentence of the Abstract states that the ACL consists of two main fiber bundles (p. 35); the body acknowledges that whether divided into two, three or multiple bundles, there is no histological septum, that a two-bundle division seems oversimplified, and that it is merely widely accepted (p. 38). Citations of this review should follow the body of the text.

Key points

  • The original double bundle was a tension-based partition: Girgis et al. (1975) described a thin anteromedial band that tightens in flexion and the remaining bulk, and recorded that the ACL is flat in extension.
  • The same names need not refer to the same fibers: Girgis called the anteromedial band the shortest part of the ACL; Steckel et al. (2007) measured AM length 37.7 mm and PL length 20.7 mm. Before comparing bundle lengths, cross-sectional areas and insertion areas across studies, the method of division in each study must be checked.
  • What is reproducible is the distribution of tension: Fibers of the anteromedial portion tighten in flexion, and those of the posterolateral portion in extension; studies reporting a double bundle and studies finding no bundling agree on this point.
  • A frequently cited review: Petersen and Zantop (2007) state in the Abstract that the ACL consists of two main fiber bundles, and acknowledge in the body that there is no histological septum and that the dichotomy is oversimplified.

3. Effects of Observation Conditions and Separation Methods on Bundle Number

Table 1 lists the principal anatomical studies discussed in this section. Studies reporting no bundling, a double bundle or a triple bundle can mostly be matched to corresponding conditions in their methods.

Table 1. Specimens, observation conditions and reported structure in the principal anatomical studies

Study Specimens Observation and separation conditions Reported structure
Girgis et al. (1975) 44 knees (20 cadaveric, 24 fresh) Manual flexion–extension to observe tension; femur split longitudinally in some knees Thin anteromedial band and posterolateral bulk; flat ligament in extension
Norwood and Cross (1979) 18 fresh amputation knees Thickened synovium had to be removed in some knees AM, IM and PL bundles
Odensten and Gillquist (1985) Gross: 20 knees; histology: 3 knees and 5 ACLs All non-collagenous tissue cleared; sections viewed under polarized light No bundling grossly or histologically
Fuss (1989) 22 knees, 20 of them separated Artificially divided into small bundles; origins and insertions traced No pre-existing septum; grouped by tension
Clark and Sidles (1990) 6 donors (64–67 years) Glutaraldehyde fixation in situ; light and scanning electron microscopy Fiber subunits and fascicles; no continuous bounding membrane for larger groups
Amis and Dawkins (1991) 27 knees Synovium removed; clefts probed with a blunt probe Triple-bundle appearance in most knees; separation in young specimens considered artificial
Mommersteeg et al. (1995) 3 ACLs (63–89 years) Three-dimensional digitization; blunt separation by fiber direction 8–9 bundle groups per ligament, partitioned differently
Steckel et al. (2007) 6 knees (3 donors, 24–40 years) Two bundles separated surgically; observed in flexion and extension AM and PL bundles in all 6 knees
Zhao and Huangfu (2010) 10 knees 90° of flexion; separation from three starting points Insertion partitions changed with the starting point
Otsubo et al. (2012) 7 knees (76–82 years) Divided into three bundles for the purpose of a triple-bundle reconstruction study AM, IM and PL bundles in all 7 knees
Śmigielski et al. (2015) 111 knees (mean 67 years) Surface membrane removed; measured in extension Flat ribbon from near the femoral end to midsubstance
Siebold et al. (2015) 20 knees (median 78 years) Flexion avoided throughout; dissected under magnification Flat midsubstance; multiple small fascicles visible in some knees
Noailles et al. (2017) 42 knees (median 80.6 years) At 90° of flexion, femur externally rotated until fibers parallel Single ribbon; incomplete proximal cleft in 11 knees
Muhammed and Jasim (2026) 19 knees (<40 years, embalmed) Blunt separation Triple bundle; AM bundle bifurcated in 7 knees
Reschenberg and Ruth (2026) 19 knees (74–109 years, embalmed) Gross judgment by 6 observers; histological judgment by 3 Gross double bundle 13/19; septum 5/19

Synovium and surface membrane

Norwood and Cross (1979) recorded in 18 fresh amputation knees that in some knees the synovium itself divided the ligament into recognizable bundles, and in others the synovium was thickened and had to be removed before the bundles could be recognized (p. 23); the original paper does not give how many knees fell into each category. Amis and Dawkins (1991), in 27 knees, found a triple-bundle appearance on inspection from within the notch in most knees. In young specimens the ACL was enveloped by synovium, the probe did not readily slip into clefts, and the anterior surface of the ligament folded into a bundle-like form in flexion. Three bundles could still be separated along the folds, which the authors regarded as artificial. In elderly specimens the individual bundles were often obvious, with only a thin synovial covering (p. 263). In the Discussion the authors wrote that separating fiber bundles with a dissector may cross the line between "demonstrating bundles" and "creating bundles" (pp. 264–265); the first conclusion is that the three bundles are "not necessarily separate entities" (p. 266). The original paper does not report in how many of the 27 knees a triple-bundle appearance was seen, nor does it give specimen ages; the statistical samples for length-change and stability testing were 9 knees (Figs. 5 and 8), and the uses of the remaining specimens among the 27 are not itemized.

Ribbon studies likewise treat careful removal of the surface membrane as a key step (Śmigielski et al., 2015), whereas Odensten and Gillquist cleared all non-collagenous tissue. There is no uniform standard among studies for how far surface tissue is removed, and the septa observed vary accordingly.

Knee angle and viewing direction

Steckel et al. (2007) observed that in extension the AM and PL bundles were nearly parallel in the sagittal plane and that in flexion they wound upon each other (p. 389). Zhao and Huangfu (2010), at 90° of flexion, could find a marking line between the two bundles from anterolateral in every specimen and from posteromedial in none (p. 185, Fig. 2). Siebold et al. (2015) avoided flexion throughout, on the grounds that flexion twists the ACL and creates the false appearance of an elliptical cross section and of bundling (p. 3137). In the same ligament, viewing from anterolateral in flexion readily shows two bundles, and viewing from medial in extension readily shows a flat band; the observation posture chosen by investigators corresponds clearly to the structure they report.

The same is true at arthroscopy. Steckel et al. (2009) compared two portals in 60 patients with an intact ACL who underwent arthroscopy for meniscal or cartilage lesions: with the arthroscope in the anterolateral portal, the two bundles could be distinguished without retracting the AM bundle in 17 cases (28.3%), and with the anteromedial portal in 40 cases (66.7%); in the remaining cases a probe was needed to retract the AM bundle before the PL bundle could be seen (p. 783, Table 1). Paired analysis is inadequately reported: observations through the two portals came from the same knees and are paired data. The Methods state only that "the chi-square test was used" (p. 783), without specifying which chi-square test or how the paired observations from the same knee were handled; the Results report only the separate proportions for each portal, without the paired fourfold counts. If McNemar's test is applied to paired data and the marginal counts in Table 1 are used to back-calculate, the difference is statistically significant no matter how the two portals are paired, and the direction of the portal effect is unchanged. When only one bundle is seen at arthroscopy, the viewing portal and whether retraction was used should be considered first.

Starting point of separation

Zhao and Huangfu (2010) directly tested the effect of artificial separation on the results. Ten fresh-frozen right knees (donor age not stated; ACLs that were too fine to be divided into more than two parts had been excluded) were fixed at 90° of flexion, and starting points were set at the proximal 1/5, 2/5 and 3/5 of the marking line between the two bundles, with blunt separation toward both ends from each. With different starting points, the direction of the dividing line on the insertions and the area occupied by the AM bundle changed: on the femoral insertion the AM bundle accounted for 38.6%, 52.2% and 65.1% of the whole insertion area, and on the tibial insertion for 25.7%, 40.4% and 60.6%, with pairwise differences all statistically significant (p. 189, Tables 2 and 4). In 8 of the 10 knees, over a segment about 5 mm long near the tibial insertion, adhesions between fibers required sharp separation (p. 188). The authors concluded that there is no natural interval to follow in the adult ACL and that dividing it into two bundles is a subjective maneuver (p. 190). The three starting points are repeated measurements on the same set of ligaments and cannot be counted as 30 independent samples; the conclusion remains clear: the insertion partitions described in anatomical studies are determined to a large extent by the path of separation.

Age and specimen preparation

The effect of age runs in different directions across studies. Amis and Dawkins held that bundling is more obvious in elderly specimens. Morales-Avalos et al. (2022), in 81 embalmed knees, could separate two bundles in every knee in the group under 50 years; in the group over 50 years (mean age 87 years in men, 82 years in women), the two bundles could not be separately recognized in 8/45 knees, and the authors also state that separation is especially difficult in elderly specimens (pp. 3404, 3407). Specimens in studies reporting no bundling had a mean age of 67 years (Śmigielski et al., 2015) or a median of 80.6 years (Noailles et al., 2017); specimens in studies reporting a triple bundle were aged 76–82 years (Otsubo et al., 2012) or all under 40 years (Muhammed and Jasim, 2026). Findings also vary within each method of specimen preparation: among fresh-frozen specimens, Śmigielski et al. and Noailles et al. reported a ribbon structure and Otsubo et al. a triple bundle; among embalmed specimens, Muhammed and Jasim reported a triple bundle, and the observers of Reschenberg and Ruth (2026) disagreed substantially on whether a double bundle was present. Age or method of specimen preparation alone cannot explain the disagreements among studies.

Interobserver agreement

Reschenberg and Ruth (2026) compared gross and histological judgments on the same set of specimens and calculated interobserver agreement. Nineteen embalmed knees came from bodies after completion of an anatomy teaching course; donors were aged 74–109 years (mean about 88 years). Six observers flexed, extended and rotated the specimens in hand and judged whether a double bundle was present; three observers judged on stitched midsubstance cross-sectional images of the ACL whether a double bundle was present and whether a connective-tissue septum existed. Results in the body of the text and Table 1 were: gross judgment of a double bundle in 13/19 (68.4%), Fleiss κ 0.198 (95% CI −0.061 to 0.458, P=0.135); histological judgment of a double bundle in 9/19 (47.4%), κ 0.438; septum present in 5/19 (26.3%), κ 0.684 (p. 3). The confidence interval of the κ for gross judgment includes 0; agreement among observers cannot be distinguished from chance agreement. The more the judgment depended on a visible tissue septum, the lower the proportion of double bundles and the higher the agreement.

The Abstract, Methods and declarations of the paper are inconsistent with the results in the body in the following respects:

  • κ values and proportions differ between places: The Abstract gives a gross κ of 0.202 and a septal proportion of 26.4%; the body and Table 1 give 0.198 and 26.3%; the body calls gross agreement "poor," whereas Table 1 and the Abstract call it "fair" (pp. 1, 3).
  • The proportion rule does not match the number of observers: The Methods define the double-bundle proportion as the proportion of specimens judged double-bundle by at least 2 of 3 observers (p. 3), yet gross assessment was performed by 6 observers, and the rule by which 13/19 was obtained is not stated.
  • The data declaration contradicts the body: The data-availability statement says that "no datasets were generated or analyzed during the current study" (p. 6), yet the body reports observations and statistics on 19 knees.

The specimens had already been used for teaching dissection, and the donors were very old; the study examined only one midsubstance cross section, observers had to choose between "double bundle" and "not double bundle," and they had been shown example images of single- and double-bundle ACLs beforehand (p. 2). These problems affect citation of specific numbers and the applicability of the results to younger populations; the main findings of low agreement on gross judgment and infrequent histological septa can still be used, with numbers taken from the body and Table 1.

Key points

  • The path of separation changes the partitions: In the same 10 knees, separation from three starting points on the marking line gave AM areas of 38.6%, 52.2% and 65.1% of the femoral insertion and 25.7%, 40.4% and 60.6% of the tibial insertion (Zhao and Huangfu, 2010). This is direct evidence that the method of separation affects the partitions observed.
  • Observation posture and viewing angle: Viewing from anterolateral in flexion readily shows two bundles; viewing from medial in extension readily shows a flat band. At arthroscopy, without retracting the AM bundle, the two bundles could be distinguished through the anterolateral portal in 28.3% and through the anteromedial portal in 66.7% (Steckel et al., 2009).
  • Low interobserver agreement: In 19 elderly embalmed knees, gross judgment of a double bundle was 13/19, κ 0.198, with a confidence interval including 0; a histological septum was seen in 5/19, κ 0.684 (Reschenberg and Ruth, 2026). In that paper the Abstract and body disagree on numbers, the proportion rule does not match the number of observers, and the data declaration contradicts the body.
  • Age and specimen preparation: The direction of results differs among studies and cannot alone explain disagreements on bundle number.

4. The Flat Ribbon Description

A flat cross section has long been recorded. Girgis et al. (1975) already stated that the ACL is flat in extension; Odensten and Gillquist (1985), in specimens of mean age about 29 years, measured the middle third of the ligament as 10±2 mm wide and 5±1 mm thick (p. 259); Clark and Sidles (1990) recorded a heart-shaped cross section. Several studies published in 2014–2017 developed this feature into the "ribbon" concept and went further in arguing that the double-bundle appearance is an artifact of twisting of a flat band.

Śmigielski et al. (2015)

This is the largest study. One hundred eleven fresh-frozen knees came from 81 donors (mean age 67 years, range 32–74 years); the Methods state that all were dissected by the first author, and the key step was careful removal of the synovium and fibrous membrane from the ligament surface, with measurements made in full extension. The authors report that in all dissected knees the ACL was a flat ribbon from about 2 mm distal to the femoral insertion to midsubstance, with no clear separation into AM and PL bundles; at 2 mm from the femoral insertion the mean width was 16.0 mm and the mean thickness 3.54 mm, and at midsubstance 11.4 mm and 3.4 mm (p. 3145).

Problems in the paper that affect citation of numbers and the denominator of the conclusions are as follows:

  • Abstract figures not found in the Results: The Abstract gives ribbon width as "11.43 to 16.18 mm" and thickness as "2.54–3.38 mm" (p. 3143); the ranges in the Results are width 12.7–18.1 mm and thickness 2–4.8 mm (at 2 mm from the femoral insertion) and width 9.8–13.8 mm and thickness 1.8–3.9 mm (midsubstance). None of the four Abstract figures appears in the Results.
  • Unclear denominator for "all specimens": The Methods state that 11 knees were excluded for severe osteoarthritis (p. 3144), yet the demographic data in Table 1 are still based on 111 knees; whether "all dissected knees" means 100 or 111 cannot be determined.
  • Imaging and histological "confirmation" lacks criteria and case-by-case results: Thirty knees were sent for CT, MRI and histology, and the Results state only that these "confirmed the above findings" (p. 3145); no CT or MRI figures or measurements are given, histology is illustrated only by an example (Fig. 4, p. 3147), and neither the judgment criteria nor case-by-case results for each examination are reported.
  • Cross-sectional areas in the Discussion have no source: Sex-specific cross-sectional areas (near the femoral end, 52 mm² in women and 55 mm² in men; midsubstance, 33 and 38 mm²) cannot be found in the Methods or Results (p. 3146).
  • The role of three physicians is not stated: The Methods and Limitations state that all dissections were performed by the first author (pp. 3144, 3148), yet the Acknowledgments thank three physicians for "active participation in the cadaveric dissections" (p. 3148). Both statements can be true—for example if the three served only as assistants—but the paper does not say which steps they took part in or whether they took part in morphological reading.

What can still be used is that, in predominantly middle-aged and elderly fresh-frozen specimens, after removal of the surface membrane and observation in extension, the ligament from near the femoral end to midsubstance was a flat ribbon, without a clear interbundle interval. The exact denominator for "no bundling in all specimens" is not stated; the imaging and histological "confirmation" lacks judgment criteria and case-by-case results and cannot serve as independent verification. No conflict-of-interest statement was found; the Discussion moves directly from anatomical observation to recommendations on graft shape and femoral tunnel position.

Siebold et al. (2015)

This study came from the same collaborative group. Of 20 knees, 6 were fresh-frozen and 14 paraffin-treated; median donor age was 78 years (62–108 years), and all were dissected by the first author under magnification. To avoid the false appearance of bundling caused by twist, the authors never flexed the knee, fixed the ligament in extension with freezing spray, transected it at midsubstance and then sectioned layer by layer toward the tibial insertion, measuring a mean midsubstance width of 9.9 mm and thickness of 3.9 mm (p. 3139, Table 2); the tibial insertion was C-shaped (insertion morphology is covered in Section 1.2).

  • The design excluded the conditions under which a bundled appearance arises: Failure to see two bundles in extension shows only that the fibers are arranged as a flat band in extension; it cannot test whether the bundling seen in flexion exists. The authors' own legend to Fig. 5 notes that several small fascicles could be recognized in some knees (p. 3140).
  • Values misplaced in Table 3: Table 3 lists the midsubstance data of "the present study" as width 11.9 mm, thickness 3.5 mm and cross-sectional area 37.0 mm² (p. 3140); the first two are in fact the measurements 5 mm from the tibial insertion in Table 2, and 37.0 mm² appears nowhere in the text or Table 2. Midsubstance values should be taken from Table 2.

A formal erratum (PMID 25149644) corrected only the order of the authors' names and is unrelated to the study data.

Noailles et al. (2017)

This study sought to unify the two descriptions. Of 60 fresh-frozen knees, 18 were excluded (ACL tear in 8, mucoid degeneration in 2, osteoarthritic changes in the notch in 7, synovial inflammatory disease in 1), leaving 42 for analysis; median age was 80.6 years, and all dissection and measurement were performed by the first author. At 90° of flexion the femur was externally rotated relative to the tibia until the fibers lay parallel and the ligament appeared flat; the required rotation averaged 83.6°±9.4°. An incomplete intraligamentous cleft in the proximal third was present in 11 of 42 knees (26%) (p. 1704). The authors wrote that if this cleft were artificially lengthened, a two-bundle ACL would result (p. 1707).

  • The end point of rotation was a single person's subjective judgment: Whether the ligament had reached a "flat appearance" was judged by the same person; rotation angle was measured with a goniometer, without a reproducibility check.
  • Odensten's conclusion is cited in reverse: The Discussion states that "although Odensten et al. described the ACL with two or three bundles" (p. 1705); the cited literature includes Odensten and Gillquist (1985), which explicitly reported no subdivision whatever.
  • The embryological argument cites the wrong paper: The authors claim that Ferretti et al. showed parallel arrangement of fetal ACL fibers and hypothesize that the ACL twists with embryonic rotation of the lower limb (p. 1707). The corresponding reference 36 is a study by Ferretti et al. of bony landmarks at the femoral insertion; the fetal study the authors actually describe (Ferretti et al., 2007) recorded parallel bundles in extension, but its main finding was a vascularized connective-tissue septum between the two bundles, and the paper contains no hypothesis of lower-limb rotation. The twist hypothesis in this argument has no literature support.
  • Clinical recommendations go beyond the study: The "Clinical Relevance" paragraph of the Abstract recommends choosing a flat rather than a cylindrical graft; a cadaveric morphological study cannot answer that question. The paper discloses that Boisrenoult and Pujol received support from Zimmer Biomet and Smith & Nephew, and Noailles from the French Arthroscopy Society (p. 1703).

What can still be used are two observations: at 90° of flexion the ACL fibers as a whole twisted by a mean of about 84°, close to the estimate of about 90° by Odensten and Gillquist; and about one quarter of elderly specimens had an incomplete proximal cleft.

Hohmann (2017), in a contemporaneous commentary, held that the flat-and-twisted view had already been described in the literature and that what Noailles et al. added was measurement of the derotation angle; he considered that the cleft could not be explained by advanced specimen age and regarded the fetal septum of Ferretti et al. as the strongest rebuttal of the ribbon view, concluding that the evidence is mutually conflicting and that current gross anatomical understanding has not yet produced clinically meaningful improvements in graft selection or surgical technique (pp. 1710–1711).

Ribbon width, age and the relationship between the two descriptions

The widths measured in the three ribbon studies differ greatly: Noailles measured 5.9 mm wide and 2.7 mm thick at 90° of flexion, 1.5 cm from the tibial insertion (p. 1704); Siebold measured 9.9 mm wide and 3.9 mm thick at midsubstance in extension; Śmigielski measured 11.4 mm wide and 3.4 mm thick at midsubstance in extension. Measurement site, knee angle and specimen preparation differ, and these figures cannot be pooled as reference values for "ribbon width." Specimens in all three studies were mostly elderly. Morales-Avalos et al. (2022) found that the femoral insertion was flat and ribbon-like in 95%–96% of the group over 50 years and semicircular in 90%–94% of the group under 50 years (p. 3407, Table 1). These data concern insertion morphology only and are a cross-sectional comparison between age groups, but they suggest that conclusions from ribbon studies should be applied with caution to young athletic populations.

The ribbon description and a functional double bundle can coexist. Fibers of a flat band are arranged along the elongated femoral insertion; in flexion the femoral insertion turns from vertical to horizontal and the ligament twists accordingly; fibers at different positions on the insertion lie at different distances from the flexion–extension axis and therefore change length differently. Girgis et al. used precisely this mechanism to explain tightening of the anteromedial band in flexion. After Noailles et al. cut the ACL from bone and flattened it, the anterior edge averaged 34.8 mm in length and the posterior edge 23.7 mm (p. 1705); the authors also note that this accords with the description of an AM bundle longer than the PL bundle (p. 1706). Ribbon studies show that the appearance of two separate cords of the two bundles in flexion is strongly influenced by twist; they do not prove that the difference in tension between anterior and posterior fibers is an artifact.

Key points

  • Flatness and twist have long been recorded: Girgis et al. (1975) recorded that the ACL is flat in extension; at 90° of flexion the fibers as a whole twist, estimated at about 90° by Odensten and Gillquist (1985) and measured at a mean of 83.6° by Noailles et al. (2017).
  • All three ribbon studies have problems in the original papers: In Śmigielski et al. (2015) Abstract figures disagree with the Results, the denominator of "all specimens" is unclear, and imaging and histological "confirmation" lacks criteria and case-by-case results; Siebold et al. (2015) designed out flexion and misplaced values in Table 3; Noailles et al. (2017) cite Odensten's conclusion in reverse and the wrong paper for the embryological argument.
  • Ribbon-width figures cannot be pooled: 5.9 mm, 9.9 mm and 11.4 mm come from different measurement sites, knee angles and specimen preparations, and specimens were mostly elderly.
  • Twist and tension-based partitions can coexist: Twist affects the appearance of separation of the two bundles in flexion; it does not negate the difference in tension between anterior and posterior fibers.

5. The Triple-Bundle Description

Anatomical and ultrastructural studies

Norwood and Cross (1979) are the source later triple-bundle studies often cite: the anteromedial, intermediate and posterolateral bundles arranged on the tibial insertion as a triangle with its apex posterior (p. 24). Functional testing used 12 amputation specimens: the surrounding capsular ligaments were first cut to produce the various patterns of rotatory instability, and the ACL bundles were then separated longitudinally and transected one by one (p. 24). How these 12 specimens relate to the 18 knees of the anatomical sample is not stated; with the surrounding ligaments already cut and the bundles artificially separated, the results cannot be taken directly as the independent actions of each bundle in the normal knee. Amis and Dawkins (1991) adopted the three-bundle division but, as noted above, regarded the three-bundle separation in young specimens as partly artificial.

Otsubo et al. (2012), in 7 fresh-frozen knees (76–82 years), first divided the ACL into anterior and posterior parts at 90° of flexion, then found a septum between the medial and lateral sides of the anterior part near the tibial insertion and divided the anterior part into AM and IM bundles. In 5 of 7 knees the septum was visible to the naked eye; in the other 2 it appeared only after an anterior drawer force was applied to tension the ligament (p. 131). Bundle number was predetermined by the purpose of the study: in the Discussion the authors state that the number of bundles remains controversial and that the purpose of this study was to provide data for anatomical triple-bundle reconstruction, and therefore the ACL was divided into three bundles (p. 130); the same passage acknowledges that on midsubstance cross section there was no septum that could indicate bundling (p. 130). That all 7 knees yielded three bundles is not evidence that a triple bundle is generally present. The text notes that the collaborator Shino designed a triple-bundle reconstruction in 2005.

In the fetal study of Ferretti et al. (2007), a second vascularized septum also appeared within the AM bundle in 2 specimens; the authors speculated that an intermediate bundle might exist, but no clear intermediate bundle was seen on gross dissection in any specimen (pp. 280–282).

Suzuki et al. (2015), from the same group as Otsubo et al., compared collagen fibrils in the midsubstance of the three bundles by transmission electron microscopy in 6 cadavers (76–92 years), renaming the three bundles the anteromedial–medial (AM-M), anteromedial–lateral (AM-L) and PL bundles. Fibril diameters in AM-M showed a bimodal distribution, with a mass-average diameter of 83.2±11.2 nm, larger than the 66.8±7.7 nm of the PL bundle; AM-L was 77.6±12.3 nm, with no statistically significant difference from either of the other two; fibril number per unit area, fibril area fraction, arithmetic mean diameter and median diameter did not differ among the three bundles (p. 914, Table 1). The Abstract overstates the data: it claims that the three bundles "have different ultrastructures" (p. 910), whereas the data support only the single difference in mass-average diameter between AM-M and PL; AM-L did not differ statistically from AM-M on any measured index. The inference in the Discussion that "AM-M is the strongest, AM-L next" (p. 915) is drawn from mass-average diameter and from prior mechanical testing by Butler et al.; the present study performed no mechanical testing of its own.

Imaging studies

The proportions of a triple bundle reported in imaging studies differ widely. MacKay et al. (2014) reviewed 73 consecutive clinical knee MRIs (ACL injury and mucoid degeneration excluded), read on 5 mm axial fat-suppressed T2-weighted images by one radiology resident with experience of ACL bundle studies, with difficult cases referred to the senior author (p. 1098). A triple bundle was counted only if displayed on more than one slice, yielding 15/73 (20.5%, 95% CI 12.9%–31.2%); another 7 knees showed a triple bundle only on the most inferior slice and were not counted; a triple bundle was seen in 3/6 examinations at 3 T and 12/67 at 1.5 T, P=0.06 (p. 1099). The Discussion overstates the data: in the Limitations the authors acknowledge that without arthroscopic, surgical or histological correlation it is impossible to determine that the imaging triple bundle is the one described in cadaveric studies (p. 1101), yet in the Discussion they write that it is reasonable to conclude that the ACL in these individuals consists of three distinct fiber bundles (p. 1100). The statement in the Limitations is accurate: 20.5% is the proportion of triple-bundle imaging under that sequence and judgment rule, not the prevalence of a triple-bundle structure.

Otsubo et al. (2017) examined 24 healthy volunteers with 3 T MRI and a 0.6 mm isotropic fat-suppressed three-dimensional sequence and reported a triple bundle in 22/24 knees (92%) (p. 280). The analysis plane was chosen by consensus of two orthopedic surgeons: on each knee a reference line was moved to select the oblique axial plane on which bundle structure was clearest (p. 279). The authors' group had proposed the AM-M, AM-L and PL anatomical division and a triple-bundle reconstruction; the paper does not state whether reading was blinded, and there was no anatomical correlation. Numerical and citation errors in the paper include the following:

  • A P value greater than 1: Comparison of triple-bundle recognition rates is written as "χ² test: P=4.110" (p. 280). A P value cannot exceed 1, and the paper gives no interpretable test result.
  • Planned agreement assessment not reported: The Methods state that two observers would outline separately to assess agreement (p. 280); no agreement data appear in the Results.
  • Sex of subjects inconsistent: The Abstract gives 14 men and 10 women (p. 276); the Methods give 15 men and 9 women (p. 278).
  • Misstatement of MacKay's numbers and definition: The Discussion writes MacKay et al. as 22 of 73 knees with a triple bundle and 53 with a double bundle (p. 281); MacKay's original is 15 triple-bundle knees (requiring more than one slice), another 7 shown on a single slice and not counted, and 51 double-bundle knees; 22 plus 53 also exceeds 73.
  • Incorrect self-citation of sample size: The Introduction states that Otsubo et al. (2012) used 6 knees (p. 277); the original used 7.
  • Odensten's conclusion used in reverse: A Limitations passage takes the "evenly distributed vascularized connective tissue between collagen fibers" seen by Odensten and Gillquist as connective tissue in interfascicular septa, to support the claim that MRI can display bundling (p. 282), whereas Odensten and Gillquist used that finding precisely to show that there was no bundling.

Under these conditions, 92% can only be understood as the proportion of knees in which the investigators, following their own triple-bundle model, could label three regions on the best plane chosen for each knee. It differs from MacKay's 20.5% in sequence, slice thickness, plane selection, readers and population studied; the difference between the two cannot be interpreted as an anatomical difference between populations.

Muhammed and Jasim (2026)

This study used blunt separation in 19 embalmed cadaveric knees (age at death under 40 years; embalmed for less than 1 month), published online ahead of print in 2025 and formally in 2026. The conclusion is that "the ACL is a three-bundle structure," with each bundle having independent femoral and tibial insertions.

  • Counting rules for bifurcation versus bundling are not stated: The first sentence of the Discussion states that the AM bundle bifurcated in 7 of 19 knees, and on the same page that specimens under 40 years showed no more than three bundles and that the ACL constantly consists of three bundles (p. 66). The paper does not explain how a bifurcated AM bundle was counted or how it was distinguished from an independent bundle; "constant three bundles" depends on this unstated rule, and bifurcation was seen in more than one third of the specimens.
  • The aging inference has no control data: Inclusion criteria limited age at death to under 40 years, and the single knee with ACL degeneration was excluded from analysis because fibers and insertions could not be examined (p. 64); yet the authors attribute more than three bundles to aging (p. 66), and the study had no older or degenerate specimens for comparison.
  • Clinical inferences are hypotheses: The text proposes that triple-bundle reconstruction may improve recovery of quadriceps strength and delay progression of osteoarthritis; the authors also acknowledge that these require further study (p. 68).

What can be used is that, in young embalmed specimens, after blunt separation by the first author, the ACL could be divided into three parts, and in some specimens the AM portion could be further subdivided.

Case reports

Two case reports show that a triple-bundle or multibundle appearance has been seen in vivo. Getzmann et al. (2020) reported a 35-year-old woman in whom 3 T MRI of both knees (slice thickness 2.5 mm) showed a triple-bundle ACL, anomalous attachment of the anterior horn of the medial meniscus to the ACL, and a medial meniscal tear (pp. 1–2); the left knee underwent arthroscopic partial meniscectomy, and the paper does not report whether the triple bundle was confirmed arthroscopically.

Gunaydin et al. (2023) reported a 40-year-old woman whose two MRIs were both read as partial ACL tear; at arthroscopy the ACL consisted of multiple small bundles without synovial covering, each tense to probing, and the authors judged it to be an intact anatomical variant (p. 559). The grounds for calling it a variant are insufficient: under anesthesia the Lachman test was grade 1, the anterior drawer grade 2 and the pivot shift grade 1, and at 1 year the Lachman remained grade 1 and the anterior drawer grade 2 (p. 559); together with a 1-year history of knee pain and 3 months of giving way, prior partial injury with loss of the synovial sheath cannot be excluded, and the paper does not discuss this. Recording of the position for the first MRI is also inconsistent: the text states full extension, whereas the legends to Figs. 1 and 2 state 30° of flexion (p. 558). Case reports can show only that a given appearance has occurred; they cannot be used to estimate its prevalence.

Key points

  • Anterior fibers can be further subdivided: In Otsubo et al. (2012) a septum between AM and IM was visible to the naked eye in 5 of 7 knees; in Ferretti et al. (2007) a second septum within the AM bundle was present in 2 fetal specimens. The triple-bundle description has an observational basis.
  • Sources of "constant three bundles": Otsubo et al. (2012) predetermined bundle number for the purpose of a triple-bundle reconstruction study; Otsubo et al. (2017) have problems including P=4.110, misstatement of MacKay's numbers and unreported planned agreement assessment; Muhammed and Jasim (2026) do not state the counting rule for AM bifurcation, and the aging inference has no control data.
  • Two MRI proportions cannot show population differences: 20.5% (MacKay et al., 2014) and 92% (Otsubo et al., 2017) come from different sequences, slice thicknesses, plane selections, readers and populations, both without anatomical correlation, and cannot support an inference of anatomical differences between populations.
  • Ultrastructure: Suzuki et al. (2015) found only the single difference in mass-average diameter between AM-M and PL; AM-L did not differ statistically from AM-M on any measured index, and the Abstract's claim of "different ultrastructures of the three bundles" overstates the data.

6. Interbundle Septa and Regional Differences

Fetal and adult interbundle septa

Ferretti et al. (2007) studied 40 fetal knees at 17–23 weeks (mean 20 weeks). All 40 were dissected under a stereomicroscope; 20 ACLs were sectioned sagittally, 10 transversely, and the remaining 10 observed only grossly (p. 279). After removal of the synovium the two bundles were readily seen; histologically, a septum of vascularized connective tissue lay between them (pp. 280–281). This is the principal histological evidence for an anatomical double bundle, and Hohmann (2017) also regarded it as the strongest rebuttal of the ribbon view. The paper also states that the septum was difficult to follow over the full length of the ACL and that a constant separation between the two bundles was confirmed only by combining serial transverse and sagittal sections (p. 280); detection of the septum was not counted: how many of the 30 histological specimens showed a septum is not given, nor is it stated whether reading was blinded. The fetal ACL was about 3.7 mm long and about 1 mm in transverse diameter, and richly vascular; the authors also note that branches of the middle genicular artery within the fetal ACL usually disappear in the adult (p. 283). Whether the fetal vascular septum is constantly retained in adults requires adult data to answer.

Adult data are inconsistent. Odensten and Gillquist (1985) saw no septum on transverse sections; Clark and Sidles (1990) saw discontinuous membranous septa without a single bounding membrane running the full length; Reschenberg and Ruth (2026) judged a septum present in 5 of 19 elderly specimens on midsubstance cross section. Yeo et al. (2025), in 15 sides from 10 unembalmed cadavers, saw on sagittal sections connective tissue linking the AM and PL bundles, with the gap widening from medial to lateral, and on micro-CT the gap between the two bundles widened at 60° of flexion (p. 3). Interbundle connective tissue can be present in the adult ACL; its presence and extent vary by individual and by site. Whether the lower frequency of adult than fetal septa results from remodeling with age is a speculation raised by Reschenberg and Ruth; longitudinal data are lacking.

Microstructure and material properties

Skelley et al. (2015) and Castile et al. (2016) used experimental data from the same 16 knees; the data of the two papers cannot be treated as two independent samples and pooled. The 16 knees came from 11 donors aged 24–53 years (5 pairs of left and right knees, 6 unpaired). Investigators first separated the AM and PL bundles with forceps according to published descriptions, then cut 3 thin strips from each bundle, 6 samples per ligament; each sample underwent preconditioning, a stress-relaxation test and a stretch-to-failure test in sequence (Skelley et al., p. 930). Skelley et al. analyzed the failure phase: toe-region modulus (7.2 vs 4.2 MPa), linear-region modulus (27.0 vs 16.1 MPa) and failure stress (2.8 vs 2.0 MPa) were all higher in the AM than in the PL bundle, and collagen alignment under load was more uniform; at the start of the failure test (after the samples had already undergone preconditioning and stress relaxation), the degree of alignment did not differ between the two bundles (pp. 932–933).

Statistical problems in the Skelley et al. paper include the following:

  • Inconsistent statistical methods: The Abstract states that comparisons used t tests (p. 928); the Methods state that Mann–Whitney tests were used because the data were non-normal (p. 931).
  • Unit of analysis and correlation not stated: The Methods state only that after exclusion of samples that failed at the clamps, each index was "averaged by bundle and compared" (p. 931), without saying at which level averaging was done or how AM and PL samples from the same knee and both knees of the same donor were handled. The Mann–Whitney test treats the two groups as independent samples; the direction of the effect of these correlations on P values cannot be judged from the paper.
  • Transition-point stress inconsistent between text and figure: The Results give transition-point stresses of 0.26 MPa for AM and 0.021 MPa for PL and call the former "slightly larger" (p. 932); in Fig. 4C the PL bar is about 0.2 MPa high (p. 931). There is no erratum and no raw data release, so the true PL value cannot be determined.

Castile et al. analyzed the stress-relaxation phase of the same experiments, using a linear mixed model to account for correlation among multiple samples from the same ligament and both knees of the same donor: peak stress (0.31 vs 0.16 MPa, P=0.049) and equilibrium stress (0.21 vs 0.10 MPa, P=0.004) were higher in the AM bundle, the magnitude of stress relaxation was smaller (−32.6% vs −35.1%, P=0.016), and collagen alignment during stress relaxation was also higher and more uniform (Results). This analysis addresses the stress-relaxation phase and did not reanalyze the moduli and failure stresses of the failure phase in Skelley et al.; the direction of results in the two phases is consistent. Funding for both studies came from the Orthopaedic Research and Education Foundation, with funds provided by the Musculoskeletal Transplant Foundation (MTF) and specimens donated by Mid-America Transplant Services (Skelley et al., p. 929). These studies compare two regions defined in advance. A difference between the two regions cannot distinguish "a clear boundary between two bundles" from "properties that change continuously from anterior to posterior"; Castile et al. also raise the latter possibility in the Introduction.

Descriptions of fiber alignment in the two regions are inconsistent across studies. Electron microscopy by Suzuki et al. (2015) held that AM-M fibrils were mostly aligned along the long axis, whereas PL fibrils were more dispersed in direction; polarized-light measurement by Skelley et al. showed more uniform AM alignment under load; light microscopy by Yeo et al. (2025) described AM fibers as a wavy or twisted nonlinear network with disordered alignment, and PL fibers as parallel and regular (pp. 3–4)—the opposite direction. Donors in Yeo et al. had a mean age of 69.1 years, and the paper has the following further problems:

  • Mechanical conclusions without measurement: From tissue morphology the paper infers that the AM bundle is the main fulcrum of knee motion and that above 60° of flexion the tensile strength of the AM bundle exceeds that of the PL bundle (pp. 4–6); none of these mechanical conclusions was measured.
  • Numbers and regional labels inconsistent: The Results give 10.33% and 5.32% as the elastic-fiber-to-collagen area ratios of the upper and lower PL regions, and also give dense-zone 13.08% and loose-zone 4.70% within the lower PL region (p. 4); the legend to Fig. 5 calls 10.33% and 5.32% the dense and loose zones (p. 6). The legend to Fig. 2 also calls Masson's trichrome, pentachrome and safranin O staining immunohistochemical stains (p. 3).
  • Values in the two panels of Fig. 5 are inconsistent: In Fig. 5A the elastic-fiber proportion in the upper AM region is about 9.5% and the elastic-fiber-to-collagen area ratio about 11.6%, matching the 9.59% and 11.36% in the text; the legend states that Fig. 5B summarizes the same four regions, yet in Fig. 5B these two upper-AM values are about 23% and 28%, and the elastic-fiber and area-ratio bars for the other regions are also markedly higher than in Fig. 5A (p. 6). The legend does not state that Fig. 5B uses another method of calculation. Fig. 5B cannot be used to verify these percentages, and the paper's materials do not show which panel reflects the original measurements.

Specimen ages, sampling sites and scales of observation differ among studies, and microscopic differences between regions cannot yet be summarized as stable "bundle features."

Key points

  • Fetal septa and adult data: Ferretti et al. (2007) saw a vascularized connective-tissue septum between the two bundles in the fetal ACL and did not report how many of 30 histological specimens showed a septum; adult data are inconsistent—Odensten and Gillquist saw no septum, Reschenberg and Ruth saw one in 5 of 19 elderly specimens, and Clark and Sidles saw discontinuous membranous septa.
  • Regional material properties differ: In the same 16 knees, the AM region had higher moduli and failure stress in the failure phase (Skelley et al., 2015; inconsistent statistical methods, unit of analysis not stated) and higher peak and equilibrium stresses with less relaxation in the stress-relaxation phase (Castile et al., 2016; linear mixed model).
  • Boundary versus gradient cannot be distinguished: Comparison of two regions cannot distinguish a clear boundary from continuous anteroposterior change. Descriptions of fiber alignment in the two regions run in different directions: electron microscopy by Suzuki et al. (2015) and polarized-light measurement by Skelley et al. (2015) hold that the AM region is more uniformly aligned, whereas light microscopy by Yeo et al. (2025) holds that the PL region is more regular; the three differ in method, specimens and scale of observation and cannot be treated as mutually contradictory results under the same conditions.

7. Implications for Clinical Interpretation

In operative notes and imaging reports, "AM bundle" and "PL bundle" refer to fibers of the anteromedial and posterolateral regions: the former tighten in flexion, the latter in extension, and there need not be a peelable bounding membrane between them. Bundle lengths, areas and insertion partitions measured in a given study depend on that study's method of division; the change in insertion partitions with the starting point of separation in Zhao and Huangfu is an example. These figures should not be applied to studies that used different methods of division, nor to individual patients.

"Triple bundle" and "interbundle gap" in MRI reports are imaging findings that depend on field strength, slice thickness, sequence and plane selection; existing studies lack anatomical correlation and cannot be used to assign an anatomical type to that knee.

Anatomical studies often propose in the Discussion recommendations on graft shape, tunnel number or tunnel position—flat grafts, triple-bundle reconstruction, C-shaped tibial insertion reconstruction and the like. These recommendations come from morphological reasoning; which reconstruction has better clinical outcomes cannot be answered by cadaveric morphological studies. Efficacy evidence is covered in the chapters on reconstruction technique in this book.

Key points

  • AM and PL are regional names: They describe the position and tension behavior of fibers and do not presuppose a peelable bounding membrane between them.
  • Measured bundle values depend on the method of division: Bundle lengths, areas and insertion partitions should not be applied directly to studies that used different methods of division, nor directly to individual patients.
  • Imaging findings and surgical recommendations: "Triple bundle" and "interbundle gap" on MRI cannot be used to assign anatomical type; recommendations on graft shape and tunnels arising from anatomical studies need to be tested by clinical research.

References

  1. [1]

    Amis AA, Dawkins GP. Functional anatomy of the anterior cruciate ligament. Fibre bundle actions related to ligament replacements and injuries. J Bone Joint Surg Br. 1991;73(2):260-267. PMID 2005151; DOI 10.1302/0301-620x.73b2.2005151.

  2. [2]

    Castile RM, Skelley NW, Babaei B, Brophy RH, Lake SP. Microstructural properties and mechanics vary between bundles of the human anterior cruciate ligament during stress-relaxation. J Biomech. 2016;49(1):87-93. PMID 26643578; DOI 10.1016/j.jbiomech.2015.11.016.

  3. [3]

    Clark JM, Sidles JA. The interrelation of fiber bundles in the anterior cruciate ligament. J Orthop Res. 1990;8(2):180-188. PMID 2303951; DOI 10.1002/jor.1100080205.

  4. [4]

    Ferretti M, Levicoff EA, Macpherson TA, Moreland MS, Cohen M, Fu FH. The fetal anterior cruciate ligament: an anatomic and histologic study. Arthroscopy. 2007;23(3):278-283. PMID 17349471; DOI 10.1016/j.arthro.2006.11.006.

  5. [5]

    Fuss FK. Anatomy of the cruciate ligaments and their function in extension and flexion of the human knee joint. Am J Anat. 1989;184(2):165-176. PMID 2712008; DOI 10.1002/aja.1001840208.

  6. [6]

    Getzmann JM, Bohnert L, Roos JE, Agten CA. Anomalous insertion of the medial meniscus into a triple bundle anterior cruciate ligament: A rare combination of two anatomical variants. Eur J Radiol Open. 2020;7:100220. PMID 32099871; DOI 10.1016/j.ejro.2020.100220.

  7. [7]

    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.

  8. [8]

    Gunaydin B, Ozcaglayan TIK, Sever C, Oznur M, Cetin MU, Can E, Eren OT. How Many Bundles Does the Anterior Cruciate Ligament Consist of? A Case Report. Sisli Etfal Hastan Tip Bul. 2023;57(4):557-562. PMID 38268645; DOI 10.14744/semb.2022.12844.

  9. [9]

    Hohmann E. Editorial Commentary: The Ribbon Theory. Another Quantum Leap? The Anterior Cruciate Ligament Is Twisted and in Fact a Flat Structure. Or not? Arthroscopy. 2017;33(9):1710-1711. PMID 28865573; DOI 10.1016/j.arthro.2017.04.011.

  10. [10]

    MacKay JW, Whitehead H, Toms AP. Radiological evidence for the triple bundle anterior cruciate ligament. Clin Anat. 2014;27(7):1097-1102. PMID 24890455; DOI 10.1002/ca.22420.

  11. [11]

    Mommersteeg TJ, Kooloos JG, Blankevoort L, Kauer JM, Huiskes R, Roeling FQ. The fibre bundle anatomy of human cruciate ligaments. J Anat. 1995;187(Pt 2):461-471. PMID 7592008.

  12. [12]

    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.

  13. [13]

    Muhammed SB, Jasim AM. Anatomy of the anterior cruciate ligament: Three-bundle concept with an apparent twist – A cadaveric study. J Arthrosc Surg Sports Med. 2026;7(1):63-69. DOI 10.25259/JASSM_38_2025; published online ahead of print 2025.09.17.

  14. [14]

    Noailles T, Boisrenoult P, Sanchez M, Beaufils P, Pujol N. Torsional Appearance of the Anterior Cruciate Ligament Explaining "Ribbon" and Double-Bundle Concepts: A Cadaver-based Study. Arthroscopy. 2017;33(9):1703-1709. PMID 28865572; DOI 10.1016/j.arthro.2017.03.019.

  15. [15]

    Norwood LA, Cross MJ. Anterior cruciate ligament: functional anatomy of its bundles in rotatory instabilities. Am J Sports Med. 1979;7(1):23-26. PMID 420384; DOI 10.1177/036354657900700106.

  16. [16]

    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.

  17. [17]

    Otsubo H, Shino K, Suzuki D, Kamiya T, Suzuki T, Watanabe K, Fujimiya M, Iwahashi T, Yamashita T. The arrangement and the attachment areas of three ACL bundles. Knee Surg Sports Traumatol Arthrosc. 2012;20(1):127-134. PMID 21695467; DOI 10.1007/s00167-011-1576-z.

  18. [18]

    Otsubo H, Akatsuka Y, Takashima H, Suzuki T, Suzuki D, Kamiya T, Ikeda Y, Matsumura T, Yamashita T, Shino K. MRI depiction and 3D visualization of three anterior cruciate ligament bundles. Clin Anat. 2017;30(2):276-283. PMID 27888604; DOI 10.1002/ca.22810.

  19. [19]

    Parrilli A, Grassi A, Orellana F, Lolli R, Marchiori G, Berni M, Fini M, Lopomo NF, Zaffagnini S. 3D visualization of the human anterior cruciate ligament combining micro-CT and histological analysis. Surg Radiol Anat. 2024;46(2):249-258. PMID 38265490; DOI 10.1007/s00276-023-03295-5.

  20. [20]

    Petersen W, Zantop T. Anatomy of the anterior cruciate ligament with regard to its two bundles. Clin Orthop Relat Res. 2007;454:35-47. PMID 17075382; DOI 10.1097/blo.0b013e31802b4a59.

  21. [21]

    Reschenberg E, Ruth AA. Interobserver variability in assessing anterior cruciate ligament anatomy: comparing gross and microanatomical observations. Surg Radiol Anat. 2026;48(1):56. PMID 41649559; DOI 10.1007/s00276-026-03826-w.

  22. [22]

    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. Erratum: Knee Surg Sports Traumatol Arthrosc. 2016;24(9):3046, PMID 25149644, DOI 10.1007/s00167-014-3230-z; corrects only the order of authors' names.

  23. [23]

    Skelley NW, Castile RM, York TE, Gruev V, Lake SP, Brophy RH. Differences in the microstructural properties of the anteromedial and posterolateral bundles of the anterior cruciate ligament. Am J Sports Med. 2015;43(4):928-936. PMID 25634908; DOI 10.1177/0363546514566192.

  24. [24]

    Śmigielski R, Zdanowicz U, Drwięga M, Ciszek B, Ciszkowska-Łysoń B, Siebold R. Ribbon like appearance of the midsubstance fibres of the anterior cruciate ligament close to its femoral insertion site: a cadaveric study including 111 knees. Knee Surg Sports Traumatol Arthrosc. 2015;23(11):3143-3150. PMID 24972997; DOI 10.1007/s00167-014-3146-7.

  25. [25]

    Steckel H, Starman JS, Baums MH, Klinger HM, Schultz W, Fu FH. Anatomy of the anterior cruciate ligament double bundle structure: a macroscopic evaluation. Scand J Med Sci Sports. 2007;17(4):387-392. PMID 16805782; DOI 10.1111/j.1600-0838.2006.00579.x.

  26. [26]

    Steckel H, Fu FH, Baums MH, Klinger HM. Arthroscopic evaluation of the ACL double bundle structure. Knee Surg Sports Traumatol Arthrosc. 2009;17(7):782-785. PMID 19333575; DOI 10.1007/s00167-009-0783-3.

  27. [27]

    Suzuki D, Otsubo H, Watanabe T, Kamiya T, Nagoya S, Yamashita T, Shino K. Ultrastructure of the three anterior cruciate ligament bundles. Clin Anat. 2015;28(7):910-916. PMID 26118465; DOI 10.1002/ca.22586.

  28. [28]

    Yeo IS, Hong JE, Yang HM. Histomorphometric analysis of anterior cruciate ligament bundles and anatomical insights into injury mechanisms. Sci Rep. 2025;15(1):6762. PMID 40000733; DOI 10.1038/s41598-025-88037-w.

  29. [29]

    Zhao J, Huangfu X. Arbitrary starting point of separation affects morphology of the 2 bundles of anterior cruciate ligament at insertion sites. Arthroscopy. 2010;26(2):184-191. PMID 20141981; DOI 10.1016/j.arthro.2009.07.012.