KneeDaily

8.2.1

Anatomic Localization of the ACL Femoral Attachment

Yongtao Maov1.0Published 29 references
Evidence through 2026.10.03Chinese original
Contents

The femoral attachment of the anterior cruciate ligament (ACL) lies on the posterior part of the lateral wall of the intercondylar notch. In the arthroscopic view with the knee flexed 90°, the dense direct insertion is a narrow band immediately on the low side of the lateral intercondylar ridge. Low to it lie the looser indirect insertion and the fan-like extension, some fibers of which continue to the posterior articular cartilage margin. The lateral bifurcate ridge divides the attachment into a deep AM area and a shallow PL area. When single-bundle reconstruction takes the center of the whole native attachment as its initial target and both ridges can be identified, the center lies about 6 mm on the low side of the lateral intercondylar ridge and about 2 mm deep to the bifurcate ridge. When the ridges cannot be identified, the starting point is 3 mm high and 12 mm shallow from the apex of the deep cartilage. After surgery, the aperture center is read on three-dimensional CT in a strict lateral projection, using a grid that covers the whole lateral condyle. With a single-bundle reference point at about 28%–30% depth and about 35% height, the direction and size of any offset are reported, together with the relation of the aperture to the lateral intercondylar ridge. A tunnel whose center lies close to the reference point, whose whole aperture lies on the low side of an identifiable ridge and which shows no marked offset in other directions can be judged close to the anatomic position. A tunnel clearly higher than the reference point, with its rim crossing the ridge, can be called high without further information.

In this chapter, "native attachment" means the whole macroscopically visible attachment, including the fan-like extension; "direct insertion" means the dense fiber zone attached to bone through a fibrocartilaginous transition, as defined histologically; and "aperture center" means the center of the intra-articular bony aperture of the femoral tunnel on postoperative imaging. Arthroscopic directions refer to the medial surface of the lateral condyle viewed through the anteromedial portal at 90° of flexion: high and low correspond to anatomic anterior and posterior in extension, and deep and shallow to proximal and distal. They are judged against the femoral axis, not against the edges of the screen. On CT, depth is measured from the deep end of the grid and height from the notch roof, both as percentages of the full grid length or height.

1. The Normal Femoral Attachment

Direct insertion, indirect insertion and posterior fibers

Both histological studies place the anterior margin of the direct insertion at the lateral intercondylar ridge, which forms the upper margin of the attachment in the arthroscopic view. Iwahashi et al. (2010) examined 8 knees from 4 donors (mean age 77 years). Dense fibers passed into bone through uncalcified and calcified fibrocartilage within a crescent-shaped depression behind the lateral intercondylar ridge; at three section levels the insertion was 7.8–9.5 mm wide, and the depression on three-dimensional CT measured 17.4±0.9 mm in length and 8.0±0.5 mm in width (pp. S15–S19, Tables 1 and 3). The surrounding loose fibers had no fibrocartilaginous transition. They reached posteriorly to the articular cartilage margin, and a small number crossed the ridge to attach anterior to it (pp. S15–S17). Sasaki et al. (2012) measured a direct insertion 4.0–6.4 mm wide at four levels in 20 knees (mean age 69.8 years). Its posterior margin lay 3.8–4.9 mm from the posterior cartilage margin, and the space between was occupied by a membrane-like indirect insertion continuous with the cartilage; the distance from the lateral intercondylar ridge to the posterior cartilage margin was 8.5–11.7 mm (pp. 1138–1142, Table 2). Both sets of widths are means for individual section levels and cannot be combined into an individual normal range. The summary row of Sasaki et al.'s table (5.3±1.1, 4.4±0.5 and 10.1±1.3 mm) gives SDs similar to the spread among the four level means, and the level at which they were calculated is not stated, so these SDs cannot be read as variation among the 20 knees either.

The two studies disagree on the posterior limit of the direct insertion. In their Abstract and Conclusion, Iwahashi et al. describe the direct insertion as reaching the posterior cartilage margin, while the legend of Fig. 3 mentions a band of loose fibers posteriorly in contact with the cartilage margin, whose width was not measured (p. S17). Sasaki et al. measured an indirect insertion zone of about 4 mm between the direct insertion and the cartilage margin, and the two groups did not use identical criteria for fibrocartilage. In the 10 younger specimens of Moulton et al. (2017; median age 56.5 years), the center of the attachment always showed a four-layer direct insertion and the most posterior part always showed a two-layer structure without fibrocartilage, with some fibers growing into the posterior cartilage margin; the macroscopic fan-like extension lay entirely behind the lateral intercondylar ridge (pp. 167–169). The posterior indirect insertion and fan-like fibers are loosely attached, yet they are ACL tissue. "Fan-like extension" describes a macroscopic appearance, and the region each study calls by that name does not correspond exactly to the histological indirect insertion.

Where a macroscopic outline ends depends on the tissue retained and on the drawing method. In the 7 knees of Colombet et al. (2006), the posterior end of the macroscopic attachment lay 0.8–3.7 mm from the adjacent articular surface, mean 2.5±1.1 mm (p. 986; p. 989, Table 2). Kaseta et al. (2008) measured 2.6±0.7 mm from the margin of the attachment to the posterior cartilage margin in 12 knees (Results). When Sasaki et al. retained the superficial fibrous membrane, the macroscopic posterior margin lay a mean of 7.8 mm from the cartilage margin, anterior to the histological direct insertion (pp. 1138, 1140, Table 1). The posterior margin of the direct insertion, the posterior margin of the macroscopic attachment and the end of the posterior fibers are three different structures. The posterior cartilage margin is a stable low-side reference during surgery; where the attachment actually ends in a given knee is shown only by an intact remnant outline that can be clearly identified.

Size and shape of the attachment

Ferretti et al. (2007) outlined the whole attachment, including the surface fibrous membrane, in 16 cadaveric knees: length 17.2±1.2 mm, width 9.9±0.8 mm at the junction of the two bundles, and three-dimensional surface area 196.8±23.1 mm² (p. 1221, Table 1; p. 1223). Kaseta et al. found that the attachment was not a regular ellipse. It widened from deep to shallow, with a maximum of 11.6 mm in the high–low direction and 16.8 mm in the deep–shallow direction (Results). In the arthroscopic view the long axis of the attachment runs roughly deep to shallow, with the AM attachment deep and somewhat high and the PL attachment shallow and somewhat low. The 17.4 by 8.0 mm CT depression of Iwahashi et al. corresponds to the narrow area occupied by the direct insertion and is narrower than the whole attachment, which includes the indirect insertion.

Lateral intercondylar ridge and lateral bifurcate ridge

The lateral intercondylar ridge is the anterior margin of the dense attachment and lies above the attachment in the arthroscopic view. Iwahashi et al. saw a small number of loose fibers crossing it, so ACL fibers are not entirely absent anterior to the ridge. Ferretti et al. found the ridge arthroscopically in all 60 patients with ACL injury and in all 16 cadaveric knees; in 1 knee it extended only to the middle of the attachment (pp. 1220–1222). In the specimens of Ziegler et al. (2011), the ridge was 18.0 mm long, and its prominence ranged from clearly visible to only slightly palpable (p. 746).

The lateral bifurcate ridge runs roughly perpendicular to the lateral intercondylar ridge, from it toward the posterior cartilage margin, and separates the AM and PL attachments. Ferretti et al. found it after clearing the stump in 49 of 60 patients and 13 of 16 cadaveric knees; it was 3.5±0.8 mm long and usually separated only the anterior part of the two bundle attachments (pp. 1220–1222, Table 1). In 11 of the 12 knees of Ziegler et al., the bifurcate ridge could be identified only after soft tissue had been removed, and in most it appeared as a change in slope between the plateau of the AM attachment and the depression of the PL attachment; the authors included this transition and measured a length of 11.6 mm (p. 746). The two lengths refer to different structures. When no bifurcate ridge is visible, the knee still has two bundle areas. The junction of the two ridges lies at the anterior margin of the attachment, and the center of the attachment lies low and deep to the junction. The 60 patients of Ferretti et al. were examined by one senior surgeon after the stump had been cleared. In the 90° examples of Fig. 4, the anterolateral portal mainly shows the PL attachment and the bifurcate ridge, whereas the anteromedial portal shows both bundle attachments, both ridges and the change of slope between them; the text states that the bifurcate ridge is best seen from the anteromedial portal (p. 1221).

The bony prominence at the posterior edge of the direct insertion described by Sasaki et al. is a different structure. The paper reports neither its prevalence nor its arthroscopic or imaging visibility; it should not be confused with the bifurcate ridge or used as a routine landmark (pp. 1142–1144).

Definitions of the attachment center

At least five different points are called the "center" of the attachment: the area center of the whole macroscopic attachment, the center of the direct insertion, the centers of the AM and PL bundles, the midpoint between the two bundle centers, and an overall center whose method of calculation is not stated. These points need not coincide. Kaseta et al. recorded the attachment surface by three-dimensional digitization and calculated its geometric center (Methods). Colombet et al. defined each bundle center as the parallel projection of the central fibers of the bundle onto the attachment surface, placing the AM center at 26.4% depth and 25.3% height and the PL center at 32.3% and 47.6%; the authors note that this differs from the geometric center of the attachment area (pp. 986, 988–989). Their Tables 1 and 2 label the dispersion of single measurements by 6 surgeons as "Intraobserver error", although the design included no repeated measurements by the same observer. The authors disclose consulting relationships with Smith & Nephew Endoscopy, and the second author's fellowship was funded by the company (p. 984). Edwards et al. (2008) located the two bundle centers on photographs by fitting ellipses to the attachment areas (p. 31). Ziegler et al. defined the bundle centers by tensioning the bundle remnants and observing the projection of the central fibers onto bone; their method for the overall center is not described separately (p. 746). Hart et al. (2015) used the midpoint between the AM and PL bundles (p. 1778). In their pooled analysis, Parkar et al. (2017) took the mean of the two bundle centers as the overall center when an original study reported only bundle centers (author's archived manuscript, p. 5).

When the outline changes, the center moves. Iriuchishima et al. (2016) compared the bundle centers in 14 knees with the fan-like extension included and excluded. Excluding it moved the AM center from 29% to 35% in depth and from 37% to 36% in height, and the PL center from 37% to 43% in depth and from 73% to 69% in height (pp. 255–256, Table 2). The "modified quadrant method" used in that study is not defined, and two outlines drawn on the same knees were compared with tests for independent samples, so only the direction of change should be used: including the fan-like extension moves both bundle centers deeper and moves the PL center lower. The study did not compare function and gives no reason to regard positions nearer the direct insertion as better. What it shows is that a remnant must be judged complete before its center is used. Coordinates also need a common origin. Edwards et al. report height from the high end in the Abstract (AM 55%, PL 62%) and from the low end in the text (45% and 38%) (pp. 29, 33). The two sets are complements; mixing the origins would reverse the relative heights of AM and PL.

In this chapter, single-bundle reconstruction takes the center of the whole native attachment as its initial target, and double-bundle reconstruction takes the AM and PL centers. The basis is that Ziegler et al. measured the distances from these three centers to both ridges and to the cartilage margins, and that Kaseta et al. tested the accuracy of visually targeting the whole-attachment center in knees with an intact attachment. None of the histological studies cited here compared clinical results after reconstruction at different centers, and the whole-attachment center has not been shown to be superior to the direct-insertion center. If a single-bundle tunnel is deliberately moved toward the AM bundle or the direct insertion, the target has changed. It should be described as such, and the distances for the whole-attachment center no longer apply.

Distances from the center to the ridges and cartilage margins

Ziegler et al. used 12 fresh-frozen knees (8 male, 4 female; mean age 57.8 years, range 45–70). After open exposure, the bundles were separated by fiber orientation and tension during flexion and extension, the coordinates of each point were recorded with an electromagnetic tracking system, and the shortest three-dimensional straight-line distance from each center to each landmark was calculated. One knee was excluded because of metallic interference, leaving 11 for quantitative analysis (pp. 744–746). Directions are given in anatomic terms with the knee extended, and Fig. 3 shows the correspondence with 90° of flexion (p. 745): posterior corresponds to low in the arthroscopic view, proximal to deep and distal to shallow.

Landmark (arthroscopic direction) Overall center AM center PL center
Lateral intercondylar ridge (center on its low side) 6.1 7.1 3.6
Lateral bifurcate ridge 1.7 (deep) 4.8 (deep) 5.2 (shallow)
Posterior cartilage margin 8.5 — 5.7
Shallow (distal) cartilage margin 14.7 18.6 10.7
Apex of the proximal curvature of the cartilage margin 12.9 11.7 16.7
Most posterior point of the cartilage margin 10.0 15.1 6.9

Values are means of 11 knees in millimeters, from p. 746 and Table 1 on p. 748. The two bundle centers were 10.0 mm apart, each 5.2 mm from the overall center. The ± values in the original table are standard errors. The 95% confidence intervals for the distances from the overall center to the two ridges, 5.0–7.2 mm and 1.2–2.1 mm, describe the precision of the mean; converted for 11 knees, the corresponding individual SDs are about 1.7 mm and 0.7 mm. All quantitative measurements were made by the first author, and agreement on repeated measurement was not reported. On p. 746 the PL attachment is first described as "distal and slightly anterior" to the AM attachment and later as "distal and slightly posterior"; this chapter uses the distances from each center to the ridges, which agree between the Results and Table 1. These issues affect precision at the millimeter level. They do not change where the center lies relative to the two ridges. The study was funded by Health South-East Norway, and the Minnesota Medical Foundation provided medical student summer research grants to 3 authors; no separate conflict-of-interest statement was found (p. 743).

These distances are shortest distances to an oblique ridge, and moving from the junction of the ridges along the horizontal and vertical axes of the screen will not reproduce them. Because the lateral intercondylar ridge runs obliquely, the PL center lies 3.6 mm from it and the AM center 7.1 mm, which does not mean that PL is higher than AM (pp. 745, 747, Figs. 4–6). The arthroscopic description of Ferretti et al. agrees: at 90° of flexion the AM bundle lies proximal to the bifurcate ridge and the PL bundle distal to it, and both lie on the low side of the lateral intercondylar ridge (p. 1224).

Kaseta et al. used cartilage landmarks. The geometric center of the attachment lay 8.6±0.8 mm from the most posterior point of the cartilage margin and 11.7±1.2 mm from the most proximal point of cartilage visible in the notch, and the margin of the attachment lay 2.6±0.7 mm and 4.5±1.0 mm from these two points (Results). Ziegler et al. measured 8.5 mm as the shortest distance to the posterior cartilage margin and 10.0 mm to its most posterior point, which they defined in extension. The reference points are defined differently in the two studies, and the values are not interchangeable. The distance from the lateral intercondylar ridge to the posterior cartilage margin varies considerably among studies: the level means of Sasaki et al. were 8.5–11.7 mm, and Bhattacharyya et al. (2018) measured a mean of 10.1 mm (range 9–11 mm) from the ridge to the midpoint of the inferior cartilage margin in 7 femora (Results 3.2). This distance determines whether the planned aperture fits between the ridge and the cartilage margin, and it has to be measured in each knee during surgery.

2. Locating the Anatomic Point Arthroscopically

Viewing conditions

Most of the arthroscopic measurements cited here were made at 90° of flexion with the lateral wall viewed face-on through the anteromedial portal; the anterolateral portal provides a second view. The bifurcate ridge usually becomes visible only after the surface soft tissue has been cleared, so the remnant outline is observed and recorded before clearing. Distances are measured with a calibrated probe, an arthroscopic ruler or a calibrated spinal needle introduced through the patellar tendon (Bird et al., 2011, p. 1262; Hart et al., p. 1780). The flexion angle for marking may differ from that for drilling: Bird et al. marked the point at 90° and then flexed the knee to 120° to drill the guidewire (p. 1262). Directions at the time of marking refer to the 90° view.

Native remnant

When the outline of the original attachment can be identified completely in a given knee (shortly after injury, with the high, low, deep and shallow margins of the remaining fibers clearly visible and consistent with the ridges and cartilage margins), mark the center of the outline and then compare it with the position given by the ridge distances. Kaseta et al. transected the intact ACL arthroscopically in 12 fresh-frozen knees (mean age 66±17 years), leaving the attachment, and examined its size and shape with a 30° arthroscope through both the anteromedial and the anterolateral portal. A femoral guide was then introduced through the anteromedial portal at 90° of flexion, and a guide pin was placed at the visually estimated center of the whole attachment. After dissection, the attachment surface was recorded with a three-dimensional digitizing stylus accurate to 0.1 mm; the guide pin lay 1.9±1.0 mm from the geometric center (Methods, Results). These were fresh transections of intact ligaments without scarring or retraction, the comparison concerned the guide pin, and the final aperture was not measured. Arthrex provided the specimens and equipment, and the authors' institution received funding from the company (Acknowledgments, footnotes). The result supports visual targeting of the center when the attachment is intact. It does not show that a point chosen only on the dense part of a remnant is equally accurate.

A remnant of which only part remains, but which can be confirmed as native tissue, shows that the original attachment was present at that site. It serves as a local check, and its center and posterior margin are not taken to represent the whole attachment. Tissue that cannot be distinguished from scar or synovium is not used for localization. Wittstein et al. (2009) found a femoral stump in 48 of 58 knees (83%) in a retrospective review and in 43 of 44 (98%) in a prospectively recorded group; patients were operated on a mean of 6.4 months after injury, 3 of them more than 1 year after injury (Results). The authors acknowledge that scar or hypertrophic synovium could be mistaken for an ACL stump, and they did not measure localization error (Discussion).

Marking the single-bundle center from the two ridges

When both ridges can be identified, measure the distance to each with a calibrated probe and place the first mark about 6 mm on the low side of the lateral intercondylar ridge and about 2 mm deep to the bifurcate ridge (Ziegler et al.: 6.1 mm and 1.7 mm for the overall center). Compare this position with the center of the remnant outline, if one can be seen. If the two differ clearly, recheck the portal, the flexion angle and the identification of the ridges, then decide on the basis of the outline that can be trusted in that knee. If the bifurcate ridge is unclear, view the wall face-on through the anteromedial portal and look for the change of slope between the AM plateau and the PL depression. If it is still unclear, change to cartilage landmarks instead of drawing an imagined bifurcate ridge.

After marking, check from the anterolateral portal that the point is not too deep or too low. Laverdiere et al. (2020) asked 12 fellowship-trained sports surgeons to mark the junction of the ridges and the intended tunnel center on three-dimensional printed bone models. The models were made from 20 patients and fixed in 90° of flexion, with no cartilage, remnant or other soft tissue, and the surgeons did not view them arthroscopically. With the native center determined on MRI as the reference, the error of the model points was 5.82±1.97 mm, compared with 3.72±2.43 mm for the tunnels actually drilled in the same patients (P = .0046); the deviation was mainly proximal and posterior, that is, deep and low in the arthroscopic view (Results). The surgeons using the models were not the 2 surgeons who operated, so the difference cannot be attributed to any single factor such as cartilage, remnant or arthroscopic view. The Abstract refers to "10 different 3D printed knees", whereas the Methods describe models made from 20 patients. The range of ridge-identification error given in the text (posterior, 2.42–8.05 mm) also disagrees with Table 2 (1.88 mm for surgeon 9), and the table should be used.

Marking AM and PL for double-bundle reconstruction

For double-bundle reconstruction, the two sets of distances of Ziegler et al. are used separately: the AM center about 5 mm deep to the bifurcate ridge and about 7 mm from the lateral intercondylar ridge; the PL center about 5 mm shallow to the bifurcate ridge, about 4 mm from the lateral intercondylar ridge and about 6 mm above the posterior cartilage margin. Edwards et al. marked clock positions around the posterior notch outlet with a clock-face disc, drew lines parallel to the femoral shaft from each clock position and measured where the lines crossed the bundle attachments (p. 31, Figs. 1 and 3a). The AM center lay at 10:30±30 min, 4.3±1.1 mm from the posterior outlet; the PL center lay at 10:00±18 min, 8.9±2.1 mm from the outlet; the greatest attachment width was 7.6±1.5 mm for AM and 6.2±2.3 mm for PL (p. 33). The two distances start from different clock positions on the outlet and cannot be treated as fixed offsets from a single point. In the same paper, some of the lower lines measured parallel to the notch roof started from the posterior condylar cartilage margin (p. 31, Fig. 3b). Once the clock positions on the outlet have been identified, these data provide a second check of depth.

A bone bridge must remain between the two tunnels, and the spacing between native bundle centers differs from the spacing planned to accommodate two tunnels. On photographs of 22 cadaveric knees, Edwards et al. measured 6.7±1.2 mm (3.6–8.9 mm) between the bundle centers. After two simulated 6 mm tunnels had been placed within the attachment with a bone bridge between them, the planned tunnel centers were 8±1.3 mm (6–12 mm) apart, with the PL tunnel about 7 mm shallower and about 5 mm lower than the AM tunnel (p. 33). The 10.0 mm spacing of Ziegler et al. is a three-dimensional straight-line distance, whereas the 6.7 mm of Edwards et al. was measured on photographs. When the bundle centers in a given knee are closer together than the sum of the two tunnel radii plus the bridge, the tunnels have to move away from the native centers, and this should be recorded.

Cartilage landmarks when the ridges are not visible

The corner of the lateral condylar cartilage margin at the deepest part of the notch provides a starting point when the ridges cannot be seen. Hart et al. called it the apex of the deep cartilage (ADC): the proximal and anterior corner of the articular cartilage margin on the medial surface of the lateral condyle, which is the deep and high corner in the arthroscopic view (p. 1778). In 8 formalin-fixed cadaveric knees (mean age 82.5 years), the medial condyle was removed, the AM and PL bundles were identified by fiber direction and by tension during flexion and extension and marked separately, and the midpoint between them was taken as the center C. With the knee at 90° and the femur parallel to the floor, "high" was measured perpendicular to the femoral shaft and "shallow" parallel to it. C lay 3 mm high (range 1–4 mm) and 12 mm shallow (11–17 mm) from the ADC, both medians; anatomically, these directions are anterior and distal (pp. 1778–1779, Fig. 2; Table 1). Kaseta et al. measured 11.7 mm from the most proximal point of cartilage to the geometric center of the attachment, and Ziegler et al. measured 12.9 mm from the apex of the proximal curvature of the cartilage margin to the overall center. The three studies define the reference point, the center and the direction of measurement differently, and the values are not interchangeable.

When the ridges are unclear but the ADC is clear, follow the method of Hart et al.: at 90° of flexion, place the first mark 3 mm high and 12 mm shallow from the ADC, viewing through the anteromedial portal and measuring with a calibrated spinal needle introduced through the patellar tendon or with a calibrated probe (p. 1780). When the ridges are visible, the ADC serves as a second reference (p. 1780). If the point obtained from the ADC disagrees with the position chosen from the ridges, first recheck the portal and the identification of the landmarks. The shallow distance ranged from 11 to 17 mm in the 8 knees of Hart et al., so individual variation can itself cause a discrepancy; in that case the identifiable ridges and cartilage margins of the knee take precedence, and the mark is not moved to fit a fixed distance.

The error statistics in Hart et al. are internally inconsistent. The Results state that, after adjustment for femoral size, the "mean absolute difference" from the measured center was 0 mm ±1.3 mm (±10%) (p. 1779), whereas Table 2 gives a median absolute difference of 1 mm with a range of 0–2 mm (p. 1780). A mean absolute difference of 0 is possible only if every error is 0, which cannot be reconciled with Table 2 or with the scatter in Fig. 3B, and the paper does not explain how ±1.3 mm and ±10% were calculated. This error figure cannot be cited. The authors used the sum of Line 2 and Line 3 in Fig. 2 as a measure of femoral size L, namely the length measured through C parallel to the femoral shaft to the shallow cartilage margin (median 22 mm, range 21–26 mm). They stated that using 3 mm and 12 mm directly when L is 18–25 mm "gives an acceptable ± 2 mm error" (p. 1780); none of the 8 knees had an L of 18–20 mm, and the statement was not verified case by case. Measurements were made by the first author with a digital caliper to the nearest millimeter, without a reliability test, and C was a known position in open dissection; the study did not compare points chosen arthroscopically with a concealed true center. The authors declare no conflicts of interest, and the acknowledgments state that P.A.M. received support from Consultation Semperform Inc (pp. 1777, 1781). What is established is the error in the error statistics and the lack of verification for an error within ±2 mm when L is 18–25 mm. The measured medians of 3 mm and 12 mm remain usable.

Kaseta et al. also described a construction that does not depend on the remnant. One line runs in the anteroposterior direction (high–low in the arthroscopic view) through the most posterior point of the cartilage margin, and another runs in the proximodistal direction (deep–shallow in the arthroscopic view) through the most proximal point of cartilage visible in the notch. Their intersection lay on average 1.8±0.9 mm posterior and 0.2±1.4 mm distal to the center of the attachment (Results). Arthroscopically, the mark is placed about 2 mm above the intersection. This intersection is a different point from the ADC; the result is the mean of 12 knees, and the authors did not validate it separately in patients.

Ruler, percentages and clock positions

In the technique of Bird et al., the tip of a 6 mm wide arthroscopic ruler is placed at 90° of flexion at the junction of the deep proximal cartilage margin and the femur, the full deep–shallow length of the lateral wall is measured, and the depth is marked at 50%. Height is set by the tunnel diameter so that about 2 mm of bone remains between the tunnel wall and the low cartilage margin, which places the center one radius plus 2 mm above the low cartilage margin (pp. 1261–1262). The authors acknowledge that the ruler sets only the depth (p. 1266). This 50% is a proportion of the full deep–shallow length of the lateral wall; it differs both from the midpoint of the long axis of the attachment and from 50% on the CT grid. Commenting on the 43% figure of Piefer et al. (2012), Spalding et al. (2014) cited the study of Bird et al. in support of the arthroscopic 50% position. Piefer et al.'s 43% was read from an idealized diagram onto which the radiographic means of 8 studies had been projected (pp. 874, 876), and Spalding et al. pointed out that the same center can lie at 43% or at 50% depending on whether it is calculated along the full bony projection or between the arthroscopic cartilage margins (pp. 538–539). Neither proportion has been validated against the native attachment of the same knee. The ruler method suits an initial mark when the cartilage outline is clear and the ridges are not, or a check of depth.

Clock positions are not used for localization. When Edwards et al. measured on photographs parallel to the femoral axis and parallel to the notch roof, the AM center moved from 10:30 to 10:00 and the PL center from 10:00 to 9:00 (p. 33). McConkey et al. (2012) asked 12 surgeons to drill 10 mm tunnels in 72 cadaveric knees (p. 2738). By Table 7, κ for the clock position of the femoral tunnel was 0.0298 among the surgeon and 3 reviewers (4 raters) and 0.0198 among the 3 reviewers; for a five-category judgment of tunnel position (ideal, too anterior, too horizontal, too posterior, too vertical), κ was 0.1345 and 0.1661 respectively (p. 2743, Table 7). At one point the text attributes 0.0198 to the 4 raters, which does not match the footnote to Table 7. The surgeons judged 92.65% (63/68) of their own femoral tunnels ideal, whereas 69.19% of 198 reviewer ratings were ideal (p. 2740, Table 1). The surgeons had seen the intact native tissue, and the reviewers saw only the drilled bone surface (p. 2744). The study shows disagreement in subjective assessment and cannot establish who was closer to the native center. It was funded by a National Institutes of Health training program and the Vanderbilt Sports Medicine research fund, and DonJoy Orthopaedics provided specimens and wet-laboratory facilities (p. 2738).

Aperture size, the ridge and the posterior bone bridge

Even when the center is close to the native center, the aperture can extend beyond the corresponding tissue, so the first mark is checked against the aperture size. The author uses two requirements: the upper rim of the aperture should not cross the lateral intercondylar ridge, and about 2 mm of bone should remain between the lower rim and the posterior cartilage margin. The 2 mm bone bridge is the operative target of Bird et al. (p. 1262) and of Pansard et al. (2015, pp. 883–884). It is a technical choice and has a different meaning from the 2.5 mm that Colombet et al. measured from the posterior margin of the macroscopic attachment to the cartilage margin. Distances and aperture size are measured in the same direction: at the mark, measure along the high–low direction to the lateral intercondylar ridge and to the posterior cartilage margin, and compare each with the actual half-width of the aperture in that direction. The distance from the center to the ridge should be at least the half-width, and the distance to the cartilage margin at least the half-width plus 2 mm. The 6.1 mm of Ziegler et al. is a shortest distance to an oblique ridge and is a different measurement from a distance along the high–low direction, so it cannot be compared directly with the half-width. An obliquely drilled aperture is elliptical, and its width in this direction is estimated from the actual aperture instead of being taken as the drill diameter (Moon et al., 2019, p. 10). After this one-directional calculation, the whole rim is still inspected directly before and after drilling to confirm that the upper rim does not cross the ridge at any point and that bone remains posteriorly.

Whether both requirements can be met depends on the ridge-to-cartilage distance along the high–low direction at the mark. If this distance is 10 mm and the aperture is 8 mm wide in this direction, the center must be at least 4 mm from the ridge and at least 6 mm from the cartilage margin, which leaves only one position, 4 mm from the ridge. With a 9 mm aperture in the same knee the two requirements cannot both be met, and the center position, the shape and size of the aperture and the bone-bridge plan all need to be reconsidered. This is an approximate calculation in one direction and does not replace inspection of the whole rim. When the distance is sufficient, the first mark chosen from the two ridges is kept; when it is not, the mark is adjusted along the high–low direction according to these requirements. Bhattacharyya et al. proposed, from the mean of 10.1 mm, that when the ridge cannot be seen the upper rim of the tunnel should be no more than 10 mm above the midpoint of the cartilage margin (Discussion). Their measured minimum was 9 mm, so under this rule the upper rim would cross the ridge in some specimens, which contradicts the authors' own requirement that the upper rim of the tunnel always remain below the ridge. The rule should not be used.

When the aperture is wider than the direct insertion, the excess can extend into the indirect fiber zone on its low side. An upper rim that does not cross the ridge and a lower rim that does not reach the cartilage margin do not in themselves guarantee that the whole rim lies within the native attachment, since, as described above, the macroscopic attachment can end several millimeters short of the cartilage margin. Whether the whole rim lies within the attachment has to be judged against the actual remnant outline of the knee.

Preoperative MRI

Preoperative MRI can provide a reference for the individual knee, and three types of image differ in meaning. Hui et al. (2016) validated the attachment center identified on routine 1.5 T MRI in 16 cadaveric knees (8 pairs) with intact ligaments. In the open group of 8 knees, the attachment was exposed and a screw was placed at its center; on postoperative MRI this anatomic center lay 2.5±0.9 mm from the center identified on preoperative MRI. In the arthroscopic group of 8 knees, localization was guided by a virtual model, and the arthroscopic tunnel center lay 4.2±2.4 mm from the preoperative MRI center; the difference between groups had P = .08 (p. 5, Table 1). Two knees in the arthroscopic group measured 5.9 and 9.2 mm, and the Abstract reports only the 3.2±0.9 mm obtained after excluding them. Both tunnels lay posterior to the attachment shown on MRI, and the authors attributed this to anatomic variation of the attachment and surgical factors (pp. 5–6). Conmed Linvatec funded the purchase of specimens (p. 2). This validation concerns intact attachments. Whether ipsilateral MRI after ACL rupture, when the stump may be retracted or scarred, can show the pre-injury attachment center has not been tested. Contralateral MRI provides a substitute reference from the same individual: Sivakumaran et al. (2021) fitted the attachment outline on contralateral three-dimensional MRI in 41 patients with ACL rupture and obtained interobserver ICCs of 0.75 and 0.85 for the center coordinates and 0.60 for area, with mean areas of 1.00 and 1.19 cm² for the two radiologists (P < .001). There was no arthroscopic or anatomic reference standard (Abstract; pp. 4–5).

Deviation with the transtibial route

Transtibial localization is constrained by the direction of the tibial tunnel, and the femoral end tends toward the front of the attachment. In same-knee comparisons by Kaseta et al., the transtibial guide pin lay 7.9±2.2 mm from the attachment center, on average 5.1±1.7 mm anterior and 3.6±3.9 mm proximal to it, that is, high and deep in the arthroscopic view and close to the anterior margin of the attachment (Results); the order of the two techniques was not randomized. Hart et al. engaged a single 7 mm transtibial offset guide (Smith & Nephew) against the posterior aspect of the lateral condyle at the level of the attachment. In none of the 8 knees did the marked point reach the midpoint between the bundles, the median distance being 4.5 mm (2–9 mm) (pp. 1778–1779). The Abstract concludes that "current commercially available transtibial femoral offset guides cannot reach the center of the ACL's femoral footprint and therefore should not be used" (p. 1777). An open-specimen experiment with one device does not support this generalization. Postoperative CT findings of the same kind from Kopf et al. (2010) are discussed in Section 3.

When different methods give different points

The remnant outline, the ridge distances, the ADC, the Kaseta construction and the ruler refer to different centers and references, and the points they give need not coincide. The author proceeds as follows. First, check whether the portal, flexion angle, reference points and chosen center are consistent. Then decide according to how trustworthy each reference is in the knee: a completely identifiable native outline first, the two ridges second, the ADC or the cartilage construction when the ridges are unclear, and the ruler as a check of depth. Group values from several methods are not averaged into one point, and the intraoperative mark is not moved toward the group reference coordinates used on postoperative CT. This order ranks the methods by how directly each refers to structures of the individual knee and is the author's clinical choice; none of the sources cited here compared the methods directly.

3. Judging Tunnel Position on Postoperative CT

Structures shown by bone-surface CT

Bone-surface three-dimensional CT shows bone: an identifiable lateral intercondylar ridge, the notch roof, the bony contour of the lateral condyle and the rim of the aperture. The articular cartilage margin and residual soft tissue are not shown, and Pansard et al. state that three-dimensional CT does not show soft tissue (p. 887). Whether a bone bridge remains between the aperture and the posterior cartilage margin, and whether the aperture covers the posterior fiber zone, cannot be examined on bone-surface CT alone; MRI, intraoperative images or records are needed as cartilage references.

Bony morphology still indicates where the main attachment lies. In their histological and CT comparison of 8 knees, Iwahashi et al. associated the direct insertion with the crescent-shaped depression behind the lateral intercondylar ridge (pp. S14–S19), so the ridge and this depression indicate the location of the main attachment. Neither the whole depression on CT, nor the subchondral bony contour, nor an estimated posterior ridge can be treated as a drawn outline of the individual direct insertion; the posterior bony prominence of Sasaki et al. was not tested for identification on CT or arthroscopy (p. 1144). The lateral intercondylar ridge is visible on CT in most specimens. Mittendorfer et al. (2026) identified it in 17 and the bifurcate ridge in 5 of 20 fresh-frozen femora from 10 older donors (63–92 years) on three-dimensional reconstruction (p. 6, Table 1). Only a discrete linear bony elevation was scored as a ridge; broad irregularities and osteophyte-like changes without a linear course were not (p. 4). CT and histology were compared only qualitatively, without quantitative registration of individual specimens (pp. 3, 5). These are detection counts in specimens, and sensitivity in postoperative patients has not been studied. A ridge not seen on CT may still be present anatomically, and its absence on CT does not indicate an abnormal attachment.

The center, the rim and coverage of the attachment are three separate judgments. A center close to the reference point means that the position is close to the chosen target; the rim shows the area occupied by the current aperture; whether the aperture lies within the native attachment of the knee cannot be answered by bone-surface CT alone. None of the sources cited here tested the accuracy of CT for judging coverage of the attachment by the aperture.

Projection correction and grid

Femoral tunnel position is measured on three-dimensional CT. The model is aligned with the distal femoral axis so that the posterior and distal contours of the medial and lateral condyles overlap in a strict lateral view; the orientation is fixed, the medial condyle is removed virtually, and measurements are made on the medial surface of the lateral condyle. Moon et al. defined the long axis by a cylinder fitted to the distal femoral metaphysis and set up coordinates on the lateral view with superimposed condyles, using the line through the lowest points of both condyles and the highest point of the posterior notch arch, before removing the medial condyle along the y–z plane (pp. 2–3). Barros et al. (2017) corrected the lateral projection by superimposing the posterior walls of both condyles in the sagittal and axial planes and the inferior walls in the sagittal and coronal planes (p. 31). The whole femur is corrected first and the viewing direction fixed before the medial condyle is removed. The cutting plane and the standard lateral image are kept, and after removal of the medial condyle the model is not rotated freely to the position where the aperture looks roundest and then treated as the lateral view. Coordinates are always measured in this fixed projection; rotating the model serves only to check the rim.

This chapter uses the Bernard–Hertel grid. Its upper border follows the notch roof (Blumensaat's line); its deep and shallow sides are perpendicular to the upper border and tangent to the deep and shallow bony contours of the lateral condyle; and its lower border is parallel to the upper border and tangent to the lowest contour of the lateral condyle. The height covers the whole lateral condyle, not the shorter height of the notch wall (Bird et al., p. 1262, Fig. 3; Kopf et al., Fig. 1A). The depth coordinate x is the distance of the aperture center from the deep end along the direction of the upper border, as a percentage of the grid length; the height coordinate y is the perpendicular distance of the center from the roof, as a percentage of the grid height. For both left and right knees the deep end is 0. Method 1 of Moon et al. belongs to the same type of grid; method 2 uses the lateral wall of the notch as the frame, and its values need their own references. On the three-dimensional model the aperture is elliptical, its shape changes with rotation, and identification of the center carries its own error (Moon et al., p. 10).

How the method changes the values

Readings can be compared with reference values only if the method is fixed. Moon et al. compared three frames on the same set of tunnels in 30 patients who had single-bundle reconstruction through the anteromedial portal by one surgeon, using CT obtained on the day of surgery (slice thickness 0.625 mm). Method 1 used the highest point of the notch as the upper border and the outer contour of the lateral condyle for the other borders; method 2 used the same upper border but the contour of the lateral notch wall for the others; method 3 used the lowest point of the notch as the upper border and the outer contour of the lateral condyle for the others (p. 3; p. 6, Fig. 5). In the neutral position the group means for height were 42.1%, 53.9% and 27.7%, and for depth 28.5%, 27.8% and 27.0% (p. 7, Table 1); changing the frame alone shifted the height group mean by 26 percentage points. Rotation of the model also changed the readings. For method 1, the depth group mean already differed significantly at 5° of varus and was 23.0% at 15° of varus and 34.5% at 15° of valgus; method 2 was least affected by rotation but showed significant height changes at 10° of internal and 15° of external rotation (p. 7, Table 1). The Abstract refers generally to "greater than 5° of rotation", yet Table 1 already shows a difference at 5° of varus, so rotation within 5° cannot be assumed to leave the readings unchanged. These are changes in group means under prescribed single-direction rotations, not upper limits of individual error, and only single-direction rotation was simulated. The frame and projection used for the reference values should be reproduced before comparison.

Part of the numerical difference among studies of the native attachment also comes from the frame. Yahagi et al. (2018) kept the attachment outline constant on true lateral photographs of 59 fixed specimens and changed only whether the upper border included the posterior hump of the roof and whether the lower and side borders followed the bony wall or the articular cartilage surface (p. 457, Fig. 2). In knees with a straight roof, changing from a bony to a cartilage frame increased the mean depth by 5.7 percentage points and reduced the mean height by 8.2 percentage points; in knees with a large hump, changing from Grid 1 to Grid 4 reduced the mean height by 16.0 percentage points (p. 458). Postoperative CT shows only bone, so the grid should follow bony contours and be compared with bony reference values. The adjusted significance level was set at P < 0.01, yet the Results also call comparisons with P = 0.027, 0.020 and 0.037 significant (pp. 457–458). These significance labels are not adopted, but the changes in the means remain usable. The authors declare no conflicts of interest and no funding (p. 460).

The method of choosing the aperture center must also be consistent. Barros et al. measured the same tunnels in 8 cadaveric knees with conventional three-dimensional rendering, in which the center was marked on the sagittal view, and with "transparent" three-dimensional CT, in which the center was marked using three orthogonal planes (p. 31). Depth values were almost identical (18.6 vs 18.3, P = 0.560), but height values differed (23.8 vs 33.0, P = 0.017) (p. 32). The paper is inconsistent about units. Fig. 3 defines t′ and h′ as percentages of the total length T and height H, with the grid marked 0–100 (p. 31), whereas the Methods and Abstract state millimeters, and Fig. 4 labels the distance between the two mean centers as 2.02 mm (p. 32). If the units were millimeters, a height difference of 9.2 could not correspond to a distance of only 2.02 mm. The values are more likely percentages, and 9.2 should not be reported as a displacement in millimeters; whatever the units, the height values of the same tunnels changed with the reconstruction and center-marking method. There was no anatomic reference standard for each knee, so these data do not show that the transparent method comes closer to the native attachment. The authors declare no potential conflicts of interest, and no separate funding statement was found (p. 30).

Measurement point and reproducibility

The measurement point is the center of the intra-articular bony aperture; the lateral cortical exit, the middle of the tunnel and the screw head are not substitutes. Kim et al. (2016) compared strict lateral radiographs and three-dimensional CT in the same 32 knees on postoperative day 7. To have a target visible on both modalities, they measured the center of the metal interference screw head at the femoral end, not the center of the tunnel aperture (pp. 1662–1663, Fig. 4). The mean coordinates were similar (X 36.3% vs 37.6%, Y 39.6% vs 41.0%), with Pearson correlation coefficients of 0.840 and 0.858; intraobserver ICCs for CT were 0.922–0.979 and interobserver ICCs 0.923–0.956 (pp. 1663–1664, Table 1). A correlation coefficient shows that the two readings vary together across patients; it does not show that the case-by-case differences are small, and the study performed no Bland–Altman analysis and reported no limits of agreement. A high ICC shows that screw-head coordinates can be reproduced by the specified method. It does not show that the screw-head center is the aperture center, and the authors acknowledge that the screw head can be eccentric within the tunnel (p. 1664). The upper border was also defined differently on the two modalities: the middle of the Blumensaat band on radiographs and the highest point of the roof on CT. From a mean band thickness of 2.83 mm and a height of 25.71 mm, the authors estimated that this could produce a difference of about 10%–11%; this is an estimate, not a remeasurement in each case (p. 1664). The series comprised 21 double-bundle and 11 single-bundle reconstructions, and 32 of 43 knees were included because strict lateral radiographs were available; the study was funded by a research grant from Chosun University (pp. 1660–1661). Radiographic readings cannot be applied directly to three-dimensional CT reference values.

In the study of Moon et al., the 95% confidence intervals of the intraobserver and interobserver ICCs were 0.95–0.98 and 0.90–0.94 (p. 5). This shows that readings are reproducible with a fixed method. It does not show that values from different methods are interchangeable or that the measured position is anatomic. Differences between observers are not negligible either: in Pansard et al. the interobserver ICCs were 0.44–0.70 and the intraobserver ICCs 0.70–0.86 (pp. 885–886).

Sources of the reference coordinates

Parkar et al. pooled 13 cadaveric studies with 218 knees and reported weighted mean femoral center coordinates of 28.6% for depth and 34.5% for height, with weighted 5th and 95th percentiles of 23.5%–37.3% and 28.4%–42.6% (author's archived manuscript, pp. 5–6, Table 2). These percentiles were calculated from study means, without the individual coordinates of the 218 knees; they describe the distribution of study means and cannot be used as an individual normal range. The suggestion in the Abstract that they be used to judge whether a postoperative tunnel is "in or out of the anatomic range" goes beyond the data. The upper limit for height also depends on an entry that does not follow the authors' own rule. The bundle heights reported by Takahashi et al. (2006) were 26.9% and 53.2%; by the authors' rule their mean should be 40.05%, but the table lists 42.6%, which is exactly the 95th percentile for height (pp. 5–6).

Bird et al. took the mean positions reported in 6 papers as their target (Table 1 gives a calculated 27.3% depth and 34.35% height, written as 28% and 35% in the text; the entry from Bernard et al. is an overall center) (p. 1263). In 50 patients treated with the ruler technique the tunnels lay on average at 30% depth and 35% height, and in 16 patients treated with a conventional offset guide at 30% and 17%. The authors report distances to the target of 5.95 and 16.17 units for the two groups without stating the units or how they were summarized (p. 1264, Fig. 6). The Abstract states that the new technique placed the tunnel "0.9 mm from the theoretic optimal center position". This figure is the difference between the mean position of the ruler group and the target, converted using the femoral dimensions of one average-sized man, and the paper states that no case-by-case measurement was made (p. 1264); it cannot be read as an error of 0.9 mm in each patient. The clinical "validation" was a comparison with literature means, not with the native attachment of each knee, and the cadaveric demonstration and arthroscopic images in the paper (pp. 1265–1266) do not supply that comparison. The conventional group consisted of patients seen more than 6 months after surgery with good outcomes, who attended for follow-up or for other reasons (p. 1261), and was selected differently from the ruler group. The authors declare no conflicts of interest; instruments from Smith & Nephew, Linvatec and ArthroCare are listed in the Methods (pp. 1259–1262).

Pansard et al. positioned tunnels using the ridges and cartilage margins. On postoperative CT in 40 patients the mean tunnel center lay at 30.4% depth and 32.3% height, close to the literature target of 28% and 35%, although depth still differed (P = .01). Individual values ranged from 12% to 53% in height and from 21% to 42% in depth (pp. 885–886), and part of this scatter comes from differences between observers. A mean close to the literature target does not show that each tunnel restored the native attachment of its knee. The authors' conclusion that "arthroscopic visualization of bony landmarks seems sufficient for accurate positioning of the femoral tunnel" goes beyond a validation against literature means. The author P.H. received consulting fees from Arthrex and Zimmer (p. 882).

Kopf et al. reviewed three-dimensional CT of 32 knees (30 patients, mean age 35.1 years) after conventional transtibial single-bundle reconstruction performed by 9 surgeons between 1989 and 2007, with CT obtained a mean of 7.6 years after surgery. Knees with reinjury or increased laxity were excluded, and 1 femur was excluded because its aperture could not be identified clearly (Materials and Methods). On the same Bernard–Hertel grid, the mean femoral aperture center lay at 37.2% depth and 11.3% height (range −5.5% to 22.9%), clustered close to or above the notch roof. The comparison AM and PL apertures came from the same group's cadaveric anatomic double-bundle reconstructions (Forsythe et al.): 21.7% and 33.2% for AM, and 35.1% and 55.3% for PL (Table II; Fig. 1A). The study was funded by the National Institutes of Health, and the authors declare no payments or benefits from a commercial entity (footnote). Because of the exclusion criteria, these data describe imaging position only; a high tunnel and clinical failure have to be judged separately.

Reference Depth x Height y Source
Single-bundle overall center (reference point used in this chapter) about 28%–30% about 35% Parkar et al., weighted mean 28.6% and 34.5%; Bird et al., target 28% and 35%, postoperative mean of the ruler group 30% and 35%
Double-bundle AM tunnel 21.7% 33.2% Kopf et al., Table II
Double-bundle PL tunnel 35.1% 55.3% Kopf et al., Table II
Conventional offset guide, 16 patients 30% 17% Bird et al., Fig. 6
Conventional transtibial, 31 knees 37.2% 11.3% Kopf et al., Table II

All of these are group means used as reference points, and they do not define an individual acceptable range. Single-bundle tunnels are compared with the single-bundle reference point, and the two tunnels of a double-bundle reconstruction with the AM and PL references respectively.

Relation of the aperture to the lateral intercondylar ridge

Magnussen et al. (2012) classified 27 femoral tunnels in patients scheduled for revision according to the relation of the current aperture to the lateral intercondylar ridge on three-dimensional CT. In type I the aperture lies entirely posterior and inferior to the ridge (low in the arthroscopic view); in type II it overlaps the ridge (high, anterior or both); in type III it lies entirely anterior and/or superior to the ridge (p. 1299, Table 1). Four views were provided for each knee (distal with 20° of external rotation, posterior, posterior with 20° of internal rotation, and the medial surface after removal of the medial condyle), and a single view could hide the aperture (pp. 1300–1302, Figs. 2–6). The CT protocol was not standardized, and when the ridge was not clearly visible, readers estimated its position from their knowledge of normal anatomy; the number of such cases and the agreement for them were not reported (pp. 1299, 1304). A majority classification, requiring agreement of at least 5 of 8 readers, was reached in 25 of 27 knees: type I in 5, type II in 9 and type III in 11 (p. 1303, Table 2).

Reported interobserver κ values lie between 0.50 and 0.57, but the paper assigns them inconsistently. The Abstract and Results give 0.57 for the three main types and 0.50 for the full classification with subtypes, whereas the all-readers column of Table 3 gives 0.50 for the main types and 0.57 for the full classification (pp. 1298, 1303, Table 3). The paper provides no basis for deciding which is correct, so it cannot be concluded that adding subtypes reduced agreement. Mean intraobserver κ values were 0.67 and 0.54, consistent between table and text (p. 1303, Table 4). The authors state that the classification reflects the current position and size of the tunnel: a tunnel whose center was originally below the ridge but which later enlarged markedly across it would also be classified as type II (p. 1304). Type I is named "well positioned", a label set by the authors and not validated against the native attachment of the knee or against arthroscopy; no funding or conflict-of-interest statement was found.

An aperture lying entirely on the low side of the lateral intercondylar ridge is consistent with the ridge forming the anterior margin of the dense attachment, but this checks only the upper side, and the type I label alone does not establish that the position is close to the anatomic attachment. Whether the aperture is too deep, too shallow or too low is still judged from the coordinates. An aperture that overlaps the ridge or lies anterior and superior to it shows that the current aperture extends beyond the anterior margin of the main attachment. When the ridge is clearly visible, the shortest distance from the aperture center to the ridge can be measured on the three-dimensional model, not on a lateral screenshot, and compared with the three-dimensional shortest straight-line distances of Ziegler et al.: 6.1 mm when the single-bundle target is the overall center, and 7.1 mm and 3.6 mm for the AM and PL tunnels of a double-bundle reconstruction.

Judging deviation and aperture enlargement

Single-bundle tunnels use the reference point of about 28%–30% depth and about 35% height. The direction and size, in percentage points, of the offset of x and y from this point are reported, together with the findings at the rim. When offsets are converted to millimeters, the actual length and height of the grid in that knee are used. All the directions below are relative to the reference point; the reference values themselves are not cut-offs for abnormality.

Close to the anatomic position: when three conditions are met together, the tunnel can be judged close to the anatomic attachment. The center lies close to the corresponding target on both axes (the reference point above for single-bundle reconstruction, the AM and PL references for double-bundle reconstruction); the lateral intercondylar ridge is reliably identifiable on CT and the whole aperture lies on its low side; and there is no marked offset toward the roof, the deep end, the shallow end or the low contour of the lateral condyle. An aperture on the low side of the ridge alone, including the type I label of Magnussen et al., is not sufficient. Whether the whole aperture lies within the native attachment of the knee is still decided from the intraoperative record of the outline.

High: the center lies on the high side of the reference point (y smaller than the reference value). A tunnel whose center is clearly higher than the reference point, whose aperture lies close to the roof and whose rim overlaps the lateral intercondylar ridge or lies anterior and superior to it is judged a high tunnel, without the need for preoperative information on the native attachment. The conventional offset-guide group and the ruler group of Bird et al. had the same depth, with heights of 17% and 35%; the conventional transtibial tunnels of Kopf et al. had a height of 11.3% and clustered close to or above the roof. Both sets of data point in the same direction as the same-knee guide-pin results of Kaseta et al.

Low: the center lies on the low side of the reference point (y larger than the reference value), and the rim approaches the low bony contour of the lateral condyle.

Deep or shallow: a center on the deep side of the reference point (x smaller than the reference value) is deep, with the rim approaching the deep bony contour; a center on the shallow side (x larger than the reference value) is shallow. In the conventional group of Kopf et al., x was 37.2%, but the main deviation was in height. Both axes must be reported, since neither axis alone shows that the position is satisfactory.

The AM and PL tunnels of a double-bundle reconstruction are compared in the same way with the AM and PL references. When the lateral intercondylar ridge is clearly visible, report whether the center and the whole aperture lie on its low side, intersect it or lie anterior and superior to it. When the ridge is not visible on CT, the roof and the bony frame of the whole lateral condyle still allow depth and height to be quantified, and a tunnel much higher than the reference point can still be judged high, although the report should not state that the rim crosses the ridge; the exact relation to the native attachment of the knee requires other information. Magnussen et al. allowed readers to estimate the ridge position, and the performance of those judgments was not tested separately. If preoperative ipsilateral MRI shows the native attachment, comparison with the postoperative tunnel is the most direct individual reference; methods for registering MRI to CT have not been validated in the sources cited here.

An aperture that crosses the ridge may result from an initial position that was too high or too anterior, or from later tunnel enlargement, and a single CT cannot distinguish between the two. The CT scans of Kopf et al. were obtained a mean of 7.6 years after surgery, and the authors discuss the possibility that enlargement moved the aperture center; Pansard et al. also list variable CT timing as a source of bias (p. 887). Later CT should take postoperative enlargement into account, and when early postoperative images are available, the two are compared to judge whether the aperture has changed.

4. Intraoperative Localization and Postoperative CT Reading

Intraoperative localization

  1. Review preoperative MRI: ipsilateral MRI that shows the native attachment serves as a reference for the knee, and contralateral MRI as a substitute reference.
  2. At 90° of flexion, view the lateral wall face-on through the anteromedial portal and check from the anterolateral portal. Observe the remnant before clearing, then clear the bone surface to identify both ridges and the change of slope.
  3. Define the target: for single-bundle reconstruction, the center of the whole native attachment as the initial target; for double-bundle reconstruction, the AM and PL centers. If the tunnel is deliberately moved toward the AM bundle or the direct insertion, record the new target.
  4. If the outline of the original attachment is completely identifiable, mark the center of the outline. A partial native remnant serves only as a local check, and tissue that cannot be distinguished from scar or synovium is not used.
  5. If both ridges are identifiable, measure with a calibrated probe. For single-bundle reconstruction, mark about 6 mm on the low side of the lateral intercondylar ridge and about 2 mm deep to the bifurcate ridge. For double-bundle reconstruction, mark AM about 5 mm deep to the bifurcate ridge and about 7 mm from the lateral intercondylar ridge, and PL about 5 mm shallow to the bifurcate ridge, about 4 mm from the lateral intercondylar ridge and about 6 mm above the posterior cartilage margin. These are shortest distances to the ridges and are not obtained by moving from the junction of the ridges along the screen axes.
  6. If the bifurcate ridge is unclear, view face-on through the anteromedial portal and look for the change of slope. If it is still unclear, or if the lateral intercondylar ridge is also unclear, mark 3 mm high (perpendicular to the femoral shaft) and 12 mm shallow (parallel to the femoral shaft) from the ADC, or about 2 mm above the intersection of the two lines of Kaseta et al.; use the 50% ruler position to check depth.
  7. Check against the aperture size. If the ridge is identifiable, measure along the high–low direction at the mark to the lateral intercondylar ridge and to the posterior cartilage margin and compare with the actual half-width of the aperture in that direction: at least the half-width to the ridge, and at least the half-width plus 2 mm to the cartilage margin. If the ridge is not identifiable, do not estimate a distance to it; leave bone in front of the visible posterior cartilage margin, check against the chosen cartilage reference and the actual aperture, and record which reference was used. If the two requirements cannot both be met, reconsider the center position, the shape and size of the aperture and the bone-bridge plan. Inspect the whole rim before and after drilling.
  8. If different methods give different points, first check the portal, flexion angle, reference points and chosen center, then decide in the order of the complete outline of the knee, the two ridges and the cartilage references. Do not average group values, and do not move the mark toward the CT reference coordinates. Check from the anterolateral portal that the point is not too deep or too low.
  9. Keep arthroscopic images and record the viewing portal, flexion angle, basis for the mark, aperture size and drilling angle for comparison after surgery.

Postoperative CT reading

  1. Record the interval between surgery and CT; use early postoperative CT where possible.
  2. On the three-dimensional model of the whole femur, superimpose the contours of the medial and lateral condyles in a strict lateral view, fix the orientation, remove the medial condyle and record the cutting plane.
  3. Place the Bernard–Hertel grid: upper border along Blumensaat's line, with the sides and the lower border tangent to the bony contours of the whole lateral condyle.
  4. After rotating the model to check the aperture outline, return to the fixed standard lateral view, mark the center of the intra-articular bony aperture and read the coordinates; do not substitute the screw head, the lateral cortical exit or the middle of the tunnel.
  5. Read x (from the deep end) and y (from the roof). Compare single-bundle tunnels with about 28%–30% depth and about 35% height, and double-bundle tunnels with AM (21.7%, 33.2%) and PL (35.1%, 55.3%). Report the direction and size of the offset in percentage points, and convert to millimeters with the grid dimensions of that knee.
  6. Describe the relation of the aperture to the lateral intercondylar ridge (entirely on its low side, overlapping it, or anterior and superior to it). When the ridge is clearly visible, measure the shortest distance from the aperture center to the ridge on the three-dimensional model and compare it with 6.1 mm for the single-bundle overall center, or with 7.1 mm for the AM tunnel and 3.6 mm for the PL tunnel.
  7. Judge the position. The tunnel is close to the anatomic position when all three conditions hold: the center lies close to the corresponding target on both axes, the lateral intercondylar ridge is reliably identifiable with the whole aperture on its low side, and no other direction shows a marked offset. It is a high tunnel when the center is clearly higher than the reference point, the aperture lies close to the roof and the rim crosses the lateral intercondylar ridge. For other directions, describe the offset from the reference point and the findings at the rim. When the ridge is not visible, judge from the roof and the remaining bony contours; a markedly high tunnel is still judged high.
  8. On later CT, consider aperture enlargement and compare with early images if available; a single CT cannot determine why an aperture crosses the ridge.
  9. Judge the relation of the aperture to the posterior cartilage margin from intraoperative images and records; bone-surface CT cannot replace them.

References

  1. [1]

    Barros MA, Fernandes TL, Dimitriou D, Pedrinelli A, Hernandez AJ. Comparison between rendering 3D-CT and transparent 3D-CT in ACL tunnel positioning. Acta Ortop Bras. 2017;25(1):30-33. PMID 28642647; DOI 10.1590/1413-785220172501167914.

  2. [2]

    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.

  3. [3]

    Bird JH, Carmont MR, Dhillon M, Smith N, Brown C, Thompson P, Spalding T. Validation of a new technique to determine midbundle femoral tunnel position in anterior cruciate ligament reconstruction using 3-dimensional computed tomography analysis. Arthroscopy. 2011;27(9):1259-1267. PMID 21741796; DOI 10.1016/j.arthro.2011.03.077.

  4. [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. [5]

    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.

  6. [6]

    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.

  7. [7]

    Hart A, Han Y, Martineau PA. The Apex of the Deep Cartilage: A Landmark and New Technique to Help Identify Femoral Tunnel Placement in Anterior Cruciate Ligament Reconstruction. Arthroscopy. 2015;31(9):1777-1783. PMID 25980920; DOI 10.1016/j.arthro.2015.03.026.

  8. [8]

    Hui C, Pi Y, Swami V, Mabee M, Jaremko JL. A Validation Study of a Novel 3-Dimensional MRI Modeling Technique to Identify the Anatomic Insertions of the Anterior Cruciate Ligament. Orthop J Sports Med. 2016;4(12):2325967116673797. PMID 28050574; DOI 10.1177/2325967116673797.

  9. [9]

    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.

  10. [10]

    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.

  11. [11]

    Kaseta MK, DeFrate LE, Charnock BL, Sullivan RT, Garrett WE. Reconstruction technique affects femoral tunnel placement in ACL reconstruction. Clin Orthop Relat Res. 2008;466(6):1467-1474. PMID 18404292; DOI 10.1007/s11999-008-0238-z.

  12. [12]

    Kim DH, Lim WB, Cho SW, Lim CW, Jo S. Reliability of 3-Dimensional Computed Tomography for Application of the Bernard Quadrant Method in Femoral Tunnel Position Evaluation After Anatomic Anterior Cruciate Ligament Reconstruction. Arthroscopy. 2016;32(8):1660-1666. PMID 27090722; DOI 10.1016/j.arthro.2016.01.043.

  13. [13]

    Kopf S, Forsythe B, Wong AK, Tashman S, Anderst W, Irrgang JJ, Fu FH. Nonanatomic tunnel position in traditional transtibial single-bundle anterior cruciate ligament reconstruction evaluated by three-dimensional computed tomography. J Bone Joint Surg Am. 2010;92(6):1427-1431. PMID 20516318; DOI 10.2106/JBJS.I.00655.

  14. [14]

    Laverdiere C, Schupbach D, Schupbach J, Harvey E, Boily M, Burman M, Martineau PA. Can Surgeons Identify ACL Femoral Ridges Landmark and Optimal Tunnel Position? A 3D Model Study. Arthrosc Sports Med Rehabil. 2020;2(4):e361-e368. PMID 32875301; DOI 10.1016/j.asmr.2020.05.008.

  15. [15]

    Magnussen RA, Debieux P, Benjamin B, Lustig S, Demey G, Servien E, Neyret P. A CT-based classification of prior ACL femoral tunnel location for planning revision ACL surgery. Knee Surg Sports Traumatol Arthrosc. 2012;20(7):1298-1306. PMID 22159522; DOI 10.1007/s00167-011-1814-4.

  16. [16]

    McConkey MO, Amendola A, Ramme AJ, Dunn WR, Flanigan DC, Britton CL; MOON Knee Group; Wolf BR, et al. Arthroscopic agreement among surgeons on anterior cruciate ligament tunnel placement. Am J Sports Med. 2012;40(12):2737-2746. PMID 23075806; DOI 10.1177/0363546512461740.

  17. [17]

    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.

  18. [18]

    Moon HS, Choi CH, Jung M, Lee DY, Chang H, Kim SH. Do Rotation and Measurement Methods Affect Reliability of Anterior Cruciate Ligament Tunnel Position on 3D Reconstructed Computed Tomography? Orthop J Sports Med. 2019;7(12):2325967119885882. PMID 31840029; DOI 10.1177/2325967119885882.

  19. [19]

    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.

  20. [20]

    Pansard E, Klouche S, Vardi G, Greeff E, Hardy P, Ferguson M. How accurate are anatomic landmarks for femoral tunnel positioning in anterior cruciate ligament reconstruction? An in vivo imaging analysis comparing both anteromedial portal and outside-in techniques. Arthroscopy. 2015;31(5):882-889. PMID 25636986; DOI 10.1016/j.arthro.2014.11.038.

  21. [21]

    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 in the text refer to the author's archived manuscript.

  22. [22]

    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.

  23. [23]

    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.

  24. [24]

    Sivakumaran T, Jaffer R, Marwan Y, Hart A, Radu A, Burman M, Martineau PA, Powell T. Reliability of Anatomic Bony Landmark Localization of the ACL Femoral Footprint Using 3D MRI. Orthop J Sports Med. 2021;9(10):23259671211042603. PMID 34692880; DOI 10.1177/23259671211042603.

  25. [25]

    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.

  26. [26]

    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. Cited only for the values entered in Table 2 of Parkar et al. (2017).

  27. [27]

    Wittstein J, Kaseta M, Sullivan R, Garrett WE. Incidence of the remnant femoral attachment of the ruptured ACL. Clin Orthop Relat Res. 2009;467(10):2691-2694. PMID 19347414; DOI 10.1007/s11999-009-0805-y.

  28. [28]

    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.

  29. [29]

    Ziegler CG, Pietrini SD, Westerhaus BD, Anderson CJ, Wijdicks CA, Johansen S, Engebretsen L, LaPrade RF. Arthroscopically pertinent landmarks for tunnel positioning in single-bundle and double-bundle anterior cruciate ligament reconstructions. Am J Sports Med. 2011;39(4):743-752. PMID 21173191; DOI 10.1177/0363546510387511.