The Gross Morphology of Torn Human Anterior CruciateLigaments in Unstable Knees

Ian K. Y. Lo, M.D., Gerardus H. R. de Maat, Jody W. Valk, and Cyril B. Frank, M.D., F.R.C.S.C.

Summary: To evaluate the presence and incidence of reattachments of tornhuman anterior cruciate ligaments (ACL), we prospectively investigated 101patients undergoing arthroscopic ACL reconstruction to study the intra-articularmorphology of ACLs under circumstances in which functional healing had failed.Results showed that roughly 72% of these unstable knees had reattachment of thetorn ACL to the posterior cruciate ligament (PCL). Eighteen percent had no signsof ACL reattachment but only 2% of previously torn ACLs were absent. Theseresults suggest that even in chronic situations in which the knee remainsfunctionally unstable, human ACLs rarely resorb. It also suggests that torn humanACLs commonly reattach in the knee, mainly to the PCL via a process that isconsistent with scarring. While the function of these reattachments is clearlyinadequate in people with unstable knees because of a combination of reattachmentlocation, scar quantity, or quality, these results nonetheless show that theintra-articular environment in humans often maintains ACL stumps and it is nottotally inhibitory to ACL reattachment via some biological process.Key Words:Anterior cruciate ligament—Ligament—Knee.

Complete tears of the anterior cruciate ligament(ACL) commonly lead to anterior instability and

knee dysfunction.1-6 Reconstruction of the ACL usingautograft or allograft tissue has become commonplacebecause of the poor functional outcomes with nonoper-ative treatment and primary repair.1-3 Poor results havebeen attributed to the poor healing potential of theACL and to the hostile intra-articular environment.This has led some authors to either state or imply thatthe ACL does not heal at all.1,7-11

During arthroscopic evaluation of an ACL disrup-tion, however, it has been noted that the ACL some-times appears to have reattached to the posterior

cruciate ligament (PCL) substance. Several authorshave recognized this and other configurations of ACLdisruptions including intrasynovial ruptures, horsehair-like tearing, bony avulsions, rounded ACL stumps, andcomplete resorption of the ligament.12-14However, fewhave prospectively documented their incidence.

In 1987, Fowler and Regan15 reported on 49 patientswith chronic ACL insufficiency whom they had treatedwith minimal arthroscopic surgery and rehabilitation.In 7 cases they noted that a portion of the ACL remnanthad reattached to the PCL. In 1991, Vahey et al.16

reported magnetic resonance imaging (MRI) findingsand correlated them to arthroscopic results. Thirtypercent of their cases involved ACL reattachment tothe PCL, which led to difficulties in MRI interpreta-tions. Others have reported on the possible function ofthis reattachment suggesting it may provide somerestraint to anterior tibial translation.17 Our purpose inthis study was to document the prevalence of this PCLreattachment versus other configurations of the ACL inpatients who had symptoms and signs of ACL defi-ciency and who were undergoing ACL reconstruction

From the Division of Orthopaedics, the University of Calgary,Calgary, Alberta, Canada (C.B.F.); the Department of Orthopae-dics, University of Western Ontario, London, Ontario, Canada(I.K.Y.L.); and the Faculty of Medicine, The University of Utrecht,The Netherlands (G.H.R.d.M, J.W.V.).

Address correspondence and reprint requests to C. B. Frank,M.D., F.R.C.S.C., Department of Surgery, the University of Cal-gary, 3330 Hospital Dr NW, Calgary, Alberta T2N 4N1, Canada.E-mail: cfrank@acs.ucalgary.ca

r 1999 by the Arthroscopy Association of North America0749-8063/99/1503-1910$3.00/0

Arthroscopy: The Journal of Arthroscopic and Related Surgery, Vol 15, No 3 (April), 1999: pp 301–306 301

at our institution. Our hypothesis was that very fewACLs would be found to be attached within the joint inthis population of patients who remained unstable andwho had thus failed to heal from a clinical point ofview.

METHODS

From the period October 1995 to October 1996, 101patients undergoing ACL reconstruction at the SportMedicine Centre of the University of Calgary werestudied. All patients had typical symptoms and signs ofACL insufficiency, with functional instability andpositive pivot shifts. There were 60 male and 41female patients with a mean age of 26.9 years (range,13 to 45 years). The right knee was involved 69 timesand the left knee 32 times. The mean time from initialinjury to reconstruction was 29.4 months (range, 1 to240 months).

All procedures were performed by four experiencedknee surgeons. Standard anteromedial and anterolat-eral portals were used for each knee arthroscopy.While viewing through each anterior portal, the grossmorphology of the ACL remnant was determined bothby observation and palpation with a probe. Theclassification of ACL attachments used by all surgeonswas a modification of that proposed by Gachter12

(Fig 1). In this classification, class A (mop-end type)was characterized by frayed or mop-like torn ends ofthe disrupted ACL. Class B was defined as an intrasy-novial or insubstance tear of the ACL. In these cases,the investing synovium appeared intact and the ACLmay appear scarred and elongated underneath. Class C

was a bony avulsion of the ACL from the tibialeminence; class D was differentiated by retracted, tornACL remnants with round or clubhead-like distention.Class E represented scarring or clear reattachment ofthe ACL to the PCL, either anteriorly, posteriorly, orboth. Class F was characterized by complete resorp-tion of the ACL with usually only a small portion of thetibial attachment recognizable. Class G representedapparent scarring of the torn ACL ends to each other;in these cases, there was no reattachment of the ACL tothe PCL and the ACL remnant appeared elongated.Patients who were classifiable into more than onegroup were categorized as class H. To minimize bias,the surgeons were not made aware of any particularpurpose to this study. Rather, they were simply askedto use the modified Gatcher classification12 to carefullyevaluate ACL morphology in every case.

In addition, patients were grouped into two broadcategories. Those with intra-articular reattachments(class B, E, G) suggesting possible scarring or healingof the ACL remnants and those without intra-articularreattachments (class A, C, D, F) suggesting the ab-sence of a scarring or healing response. Classificationwas confirmed by at least one other observer. Afterclassification, each patient underwent reconstructionby that surgeon’s preferred technique.

Data were analyzed with the assistance of a consult-ing statistician. Statistical significance was set atP ,.05. Unpairedt tests were used to compare differencesbetween sexes to define potential gender differences inthe intra-articular classification of ACL disruptionsand also to compare differences between classification,

FIGURE 1. Classification of theintra-articular morphology ofanterior cruciate ligament dis-ruptions. Note that Class H (notshown) is defined as a combina-tion of 2 or more classes.

302 I.K.Y. LO ET AL.

time from injury, and those patients with and withoutintra-articular reattachments.

RESULTS

No significant differences were found betweenobservers for the classification of ligament morphol-ogy. The overall distribution of intra-articular morphol-ogy seen at a mean of 29.4 months after initial injury(range, 1 to 240 months) from the pooled observationsis shown in Fig 2. The vast majority (66%) of caseswere class E, representing reattachment of the ACLremnant to the PCL. In 9 cases, the ACL could beclassified into more than one class (class H). In 6 ofthese 9 cases, this represented a combination of class Ewith another. Thus, combined with the pure class Eresults, there were roughly 72% of patients with somereattachments of the torn ACL to the PCL. Notsurprisingly, there were no patients in this series ofchronic reconstructions which were classified as C(bony avulsions). Such avulsions are normally treatedacutely by internal fixation in our centre.

The distribution of morphology by sex is listed inTable 1, showing no significant differences. The meantime from injury to reconstruction is also summarizedin Table 1. There was no consistent trend that corre-lated time from injury to grade or progression ofclassification (r2 5 .05). Class B and G were seenearlier after injury that other classes (P , .0001), butthe numbers are small. Patients with (class B, E, G)and without (class A, C, D, F) intra-articular reattach-

ments (class B, E, G) were seen 28.46 41.3 and33.1 6 32.8 months (mean6 SD). These intervalswere not statistically different (P 5 .63).

DISCUSSION

This study evaluated the morphology of the dis-rupted ACL in patients who had failed nonoperativetreatment and were undergoing ACL reconstruction.By definition, because all patients were undergoingsurgical reconstruction for symptoms of ACL insuffi-ciency, all of them had failed to heal functionally.Therefore, any ACL reattachments that were seen inthis series must be noted to clearly represent functionalfailures of any healing response. As such, these resultscould simply be used to support the well-knownclinical concept that ACL healing results are oftenmechanically deficient.1 Further consideration, how-ever, suggests the important distinction that this defi-ciency is not attributable to resorption of the damagedACL or to a total lack of any repair response aroundthe torn ACL.

Specifically, even in this series of unstable knees,nearly three quarters of disrupted ACLs showed someintra-articular reattachment. The majority of cases(66%) showed new attachments of the ACL to the PCLalone (class E) (Fig 3), or reattachment to the PCL plusnew attachments elsewhere. All reattachments weresubstantial enough to require the surgeon to cut orshave the reattached ACL away from the PCL, or fromits other locations in the joint. No other specific test ofthe quality or functionality of ACL attachments wasattempted. In 16% of cases (class A and D), torn endsof semiacutely torn ACLs were still free. It wasperhaps still too soon after their injury for theseparticular patients to have had their torn ACL reach amorphological endpoint; ACLs in those 16% couldthus still either resorb (class F) or attach to the PCL(class E) with time. In 7 % ofcases, loose but presentACL remnants were seen to be in continuity betweennormal ACL insertion sites (class B and G). Theselesions could have occurred as a result of incompleteACL injuries. Alternatively, however, ends of com-pletely torn ACLs could have scarred together andremained, or become (by stretching), functionallyloose. Importantly, very few (2%) ACLs in this serieshad actually resorbed, despite the fact that the vastmajority represented chronic lesions at an average of29.4 months after ACL injury.

This observation of ACL reattachments after injuryis, in itself, not unique. As noted above, other groupshave commented previously on ACL reattachments.

FIGURE 2. Distribution of disrupted anterior cruciate ligamentmorphology.

303GROSS MORPHOLOGY OF TORN HUMAN ACL

However, when data in this current study are comparedto such reports,15,16 the incidence of intra-articularreattachments documented here is higher. Fowler andRegan15 and Vahey et al.16 reported incidences of ACLreattachment to the PCL of only 14% and 30%,respectively. The difference between these reports andour study, we speculate, is not due to unique popula-tions but rather to study design. Both previous re-ports15,16 were retrospective reviews in which thesurgeon was not specifically observing and classifyingthe disrupted ACL morphology. In fact, a more recentseries of conservatively treated ACL disruptions re-ported by Ihara et al.18,19that underwent arthroscopy 3months after injury with a view toward defining ACLmorphology (more analogous to the purpose of thisstudy), found that 78% of ligaments remained ‘‘incontinuity.’’ Without evidence of complete ACL tears

and/or proof of abnormal (i.e., new) attachments oftorn ACLs in that series,18,19 it is not clear that Ihara’sresults represent true reattachments, as opposed topartial injuries. While we did not document thecompleteness of ACL injury at the original episode,our demonstration of abnormal sites of ACL reattach-ment, on the other hand, almost completely rules outthe possibility that ACL morphologies resulted frompartial ACL injuries.

What is perhaps the most interesting observation inthe current clinical series of functionally deficient,complete ACL injuries, is that very few torn ACLs hadresorbed. Instead, most had formed some sort of newintra-articular attachments. This appears to contradictthe clinical notion that ACLs do not heal functionallybecause of the implied absence of any intra-articularhealing response.7-10 In 1938, Palmer10 stated that ‘‘asa rule, total rupture of a cruciate band is probablyincapable of healing spontaneously.’’ Arnoczky7 , in1990, concluded that ‘‘although the anterior cruciateligament is capable of a vascular response after injury,spontaneous repair (or healing by second intention)does not occur.’’ In 1994, Woo et al.11 stated that‘‘midsubstance ACL tears usually do not heal’’ and Fuet al.9 suggested that ‘‘a torn anterior cruciate ligamentoften fails to show any healing response.’’ Our resultsmost clearly contradict at least the last of thesestatements in that, despite their failure to heal function-ally, some healing (reattachment) processes can and dooccur in many torn ACLs. While all reattachmentswere functionally inadequate in the series of patientswe studied (which were actually selected for surgerybased on functional failure), the presence of some ACLhealing responses is worth noting. We believe that thisis an important distinction, because no ACL healingresponse implies that reattachment is virtually impos-sible.

FIGURE 3. Arthroscopic appearance of the tibial remnant of ananterior cruciate ligament disruption reattached to roof of the notchand the posterior cruciate ligament (open arrow). The femoral endof the ACL (solid arrow) is also seen to be attached to theconvergence of the ACL stump and the PCL (Class E).

TABLE 1. Classification and Time From Injury of Disrupted ACLs by Sex

Classification

Female Male Male and Female

No. (%)MeanTSI No. (%)

MeanTSI

OverallMean TSI SD

A 4 (9.8) 54.0 3 (5.0) 16.0 37.7 30.0B 2 (4.9) 8.0 2 (3.3) 10.0 9.0 1.5C 0 (0.0) 0.0 0 (0.0) 0.0 0.0 0.0D 4 (9.8) 32.3 5 (8.3) 30.5 31.3 38.2E 26 (63.4) 25.6 41 (68.3) 32.7 29.9 42.5F 1 (2.4) 72.0 1 (1.7) 20.0 46.0 36.1G 1 (2.4) 15.0 2 (3.3) 12.0 13.0 2.0H 3 (7.3) 64.8 6 (10.0) 4.4 24.5 54.8

Abbreviations: TSI, time since injury (months); SD, standard deviation.

304 I.K.Y. LO ET AL.

Reattachment itself, of course, cannot be used toimply function, as any functionally reconnected liga-ment should perform a mechanical or physiologicalfunction (i.e., carry some tensile load). In the case ofthe ACL, the load-carrying function of intra-articularreattachments similar to those observed here, have, infact, been tested by Crain et al.20 In a study of 48patients undergoing ACL reconstruction, they docu-mented the gross morphology of the ACL remnant andmeasured KT-1000 displacement before and afterresection of ACL reattachments. They reported fourdifferent morphologies and noted differential changesin KT-1000 measurements according to ACL morphol-ogy. Patients with reattachments to the lateral oranterior intercondylar notch had the largest changes inlaxity, but patients with ACLs scarred to the PCL alsoshowed differences. This provides some evidence thatreattachents can contribute to the stability of thehuman knee.

Interestingly, these results in humans appear to bedifferent than results of ACL transections in severalanimals, supporting the concept that humans may havea more favorable biological response to ACL in-jury.21-24 In animal models, as opposed to any ACLreattachment, complete ACL resorption after an injuryis apparently very common.25-27 In the canine model,for example, O’Donoghue et al.25,26 reported that inACLs that were completely transected and treatedwithout repair, 23 of 24 ACLs retracted and resorbed.In 1 case, the ACL was reattached to the lateral portionof the intercondylar notch. More recently, Hefti et al.23

reported that 22 of 24 rabbit ACLs failed to regenerateafter complete transection. In only two animals theACL remnant reattached to the PCL. These apparentdifferences in intra-articular morphology between thesemodels and human results may represent differences instudy designs or in mechanisms of injury, but may alsoindicate differences between healing responses inanimal and human ACLs, which remain to be defined.

Finally, while we must again make it clear that allpatients in the current series clearly did not exhibit afunctional healing response, we believe that the factthat most exhibited some reattachment response is avery important distinction for two reasons. First, thissuggests that if some patients formed new attachmentswith more appropriate qualities or quantities, in moreappropriate locations within the joint, some mayactually reattach functionally without reconstruction.This could explain why some documented ACL injurypatients are able to cope without surgery.1,19 Second,while clinical data clearly show that current primaryACL repair techniques do not reliably promote func-

tional ACL healing over time1-3, results of this studysuggest that other means of promoting functionalreattachment of torn ACLs by enhancing these endog-enous biological responses should still be explored.

Acknowledgment: We thank Dr. G.D. Bell, Dr. R. Bray,and Dr. N. Mohtadi for their valuable contributions in thisstudy.

REFERENCES

1. Frank C, Jackson DW. The science of reconstruction of theanterior cruciate ligament.J Bone Joint Surg Am1997;79:1556-1576.

2. Hawkins RJ, Misamore GW, Merritt T. Follow-up of the acutenon-operative isolated anterior cruciate ligament tear.Am JSports Med1986;14:205-210.

3. Johnson RJ, Beynnon BD, Nichols CE et al. The treatment ofinjuries of the anterior cruciate ligament.J Bone Joint Surg Am1992;74:140-151.

4. Kannus P, Jarvinen M. Conservatively treated tears of theanterior cruciate ligament.J Bone Joint Surg Am1987;69:1007-1012.

5. Odensten M, Hamberg P, Nordin M et al. Surgical or conserva-tive treatment of the acutely torn anterior cruciate ligament.Clin Orthop1985;198:87-93.

6. Pattee GA, Fox JM, Del Pizzo W et al. Four to ten yearfollow-up of unreconstructed anterior cruciate ligament tears.Am J Sports Med1989;17:430-435.

7. Arnoczky SP. Basic science of anterior cruciate ligament repairand reconstruction.AAOS Instr Course Lect1990;40:201-212.

8. Frank CB. Ligament healing: current knowledge and clinicalapplications.J Am Acad Orthop Surg1996;4:74-83.

9. Fu FH, Harner CD, Johnson DL. Biomechanics of kneeligaments: Basic concepts and clinical applications.InstrCourse Lect1994;43:137-150.

10. Palmer I. On the injuries of the ligaments of the knee joint: aclinical study.Acta Chir Scand (Suppl)1938;53:1-282.

11. Woo SLY, An KN, Arnoczky SP, et al. Anatomy, biology andbiomechanics of tendon, ligament, and meniscus. In: SimonSR. ed.Orthopaedic basic science.Park Ridge, IL: AmericanAcademy of Orthopaedic Surgeons, 1994;45-88.

12. Gachter A. The various faces of anterior cruciate ligament tearsduring arthroscopic examination. In: Jakob RP, Staubli HU,eds.The knee and the cruciate ligaments.Berlin: Springer-Verlag, 1992;190-192.

13. Jackson RW, Dandy DJ. Arthroscopy of the knee., New York:Grune & Stratton, 1976.

14. Johnson LL.Diagnostic and surgical arthroscopy. St. Louis,CV Mosby, 1981.

15. Fowler PJ, Regan WD. The patient with symptomatic chronicanterior cruciate ligament insufficiency. Results of minimalarthroscopic surgery and rehabilitation.Am J Sports Med1987;15:321-325.

16. Vahey TN, Broome DR, Kowmas JK et al. Acute and chronictears of the anterior cruciate ligament: Differential features atMR imaging.Radiology1991;181:251- 253.

17. Kurzweil PR, Jackson DW. Chronic anterior cruciate liga-ments. In Fu FH, Harner CD, Vince KG, eds.Knee surgery.Vol1. Baltimore: Williams & Wilkins, 1994;731-747.

18. Ihara H, Megumi M, Takayanagi K, et al. Acute torn meniscuscombined with acute cruciate ligament injury. Second lookarthroscopy after 3-month conservative treatment.Clin Orthop1994;307:146-154.

19. Ihara H, Miwa M, Deya K, et al. MRI of anterior cruciateligament healing.J Comput Assist Tomogr 1995;20:317-321.

305GROSS MORPHOLOGY OF TORN HUMAN ACL

20. Crain EH, Fithian DC, Daniel DM. Variation in ACL scarpattern: Does this contribute to anterior stability in ACLdeficient knees? Presented at the Annual Meeting of theArthroscopy Association of North America, San Diego, CA,1997.

21. Deie M, Ochi M, Ikuta Y. High intrinsic healing potential ofhuman anterior cruciate ligament.Acta Orthop Scand1995;66:28-32.

22. Hefti F. Healing process. In Jakob RP, Staubli HU, eds.Theknee and the cruciate ligaments.Berlin: Springer-Verlag,1992;257-261.

23. Hefti FL, Cress A, Fasel J et al. Healing of the transectedanterior cruciate ligament in the rabbit.J Bone Joint Surg Am1991;73:373-383.

24. Kleiner JB, Roux RD, Amiel D, et al. Primary healing of theanterior cruciate ligament. Presented at the 32nd AnnualMeeting of the Orthopedic Research Society, New Orleans,LA, 1986.

25. O’Donoghue DH, Frank GR, Jeter WJ, et al. Repair andreconstruction of the anterior cruciate ligament in dogs.J BoneJoint Surg Am1971;53:710-718.

26. O’Donoghue DH, Rockwood CA, Frank GR et al. Repair of theanterior cruciate ligament in dogs.J Bone Joint Surg Am1966;48:503-519.

27. Woo SLY, Young EP, Ohland KJ et al. The effects of transectionof the anterior cruciate ligament on healing of the medialcollateral ligament: A biomechanical study of the knee in dogs.J Bone Joint Surg Am1990;72:382-392.

306 I.K.Y. LO ET AL.