Article Text

This article has a correction. Please see:

Download PDFPDF

Why bone–patella tendon–bone grafts should still be considered the gold standard for anterior cruciate ligament reconstruction
  1. James R Carmichael,
  2. Mervyn J Cross
  1. North Sydney Orthopaedic and Sports Medicine Centre, 286 Pacific Highway, Crows Nest, NSW 2065, Australia
  1. James R Carmichael, BMedSci, BMBS, MSc, FRCS (Tr. & Orth.), Orthopaedic Fellow, North Sydney Orthopaedic and Sports Medicine Centre, 286 Pacific Highway, Crows Nest, NSW 2065, Australia; orthojim{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

The patient population that will place the highest demand on the structural integrity and function of an ACL graft, whilst at the same time having the lowest tolerance for any donor site morbidity, is the professional sportsman or woman. The graft most appropriate for this group, regardless of technical difficulties, should therefore be considered the gold standard. We believe that the bone – patella tendon – bone (BPTB) graft best fits these criteria.

Proponents of the hamstring-based reconstruction will often cite the morbidity of the patellar tendon harvest, rather than the merits of the hamstrings, as the principal reason for avoiding the use of BPTB. Concerns exist regarding quadriceps deficit, arthritis, postoperative stiffness, anterior knee pain and sensory disturbance. The complications associated with hamstrings grafts must not be forgotten. These include increased knee laxity, hamstrings deficit and pain, tunnel widening, anterior knee pain and sensory deficit.


Graft laxity and fixation

The stability of the hamstrings reconstruction when compared with the BPTB graft is frequently questioned. It is a common finding that there is a trend to increased laxity when hamstrings grafts are used.13 This has been confirmed in a meta-analysis by Goldblatt et al, who identified a marked reduction in side-to-side laxity difference with the use of the BPTB graft.4 A similar analysis by Forster et al identified a trend to increased chance of a positive pivot jerk in the hamstrings group.5 When laxity is identified in a BPTB graft it is usually associated with tunnel misplacement.6

The cause for the increased laxity with hamstrings has been studied. In 2002 Rittmeister et al analysed the components of postoperative laxity in a cadaveric study.7 They found that the majority of graft slippage occurred at the graft bone interface, with only 8% of elongation being due to plastic graft deformation. There has been extensive review of the fixation techniques for hamstrings grafts. Tibial sided graft fixation is regarded as more problematic as all techniques rely on friction between graft, fixation device and bone in an area of relatively low bone density. Where interference screw fixation is tested to failure, the weakness appears to be in graft slippage around the screw.8 9

Femoral fixation is less problematic due to a combination of higher bone density and the possibility of suspensory fixation, which prevents graft slippage but may be associated with bungee or windscreen wiper effects which have been implicated in tunnel widening10 11 and in inhibiting bone tendon healing.12 Suspensory systems, however, result in fixation distal to the joint surface. This causes some concern as knee stability is improved with juxta-articular fixation on the tibial side13 and it would be reasonable to assume similar advantages on the femoral.

The difficulties of maintaining a secure fixation with hamstring grafts when compared with BPTB are further evidenced by the discrepancy in fixation options between the two graft types. Whereas the surgeon using the hamstrings graft is presented with a baffling array of fixation options, the vast majority of BPTB grafts are fixed with interference screws. Alternative (suspensory) systems are usually reserved for cases of graft–tunnel mismatch or posterior wall blowout.14

Tunnel widening

The phenomenon of tunnel widening has been the source of much speculation and debate. The mechanism for its occurrence is not fully understood, but what is accepted is the increased incidence of tunnel widening with the use of soft tissue grafts. Whilst of questionable clinical significance in the context of the well-functioning primary graft, tunnel widening becomes important in the revision situation as widening may compromise revision graft positioning and fixation.

Graft healing

Graft tunnel healing, as evidenced by obliteration of the bone tunnels or prevalence of tunnel widening, appears to be less problematic in BPTB grafts.3 11 15 Some basic science work using dog models does exist to support the suggestion that hamstrings graft healing to bone is slower than BPTB.16 Furthermore, it is possible to position the bone block at the femoral tunnel aperture and thus maintain a physiological graft take-off.

Return to sport

Laxdal et al in 2006, in a prospective comparison, identified a significant difference in the Tegner activity level of BPTB patients at follow-up when compared with hamstrings graft patients.17 In 2002, 97% of surgeons treating professional sportsmen from the American NFL used the BPTB graft.18 Laxity associated with hamstrings is unlikely to be the sole cause of this discrepancy. Hamstring weakness in flexion1922 and rotation23 24 is reported. Although there is compensation for the graft harvest through regeneration of the harvested tendons2527 and there may be some hypertrophy of the biceps femoris,24 the functional weakness of the hamstrings is persistent and may contribute to the reduced return to activity in this group. It is also our anecdotal experience that the use of the hamstrings graft in the elite sports player results in increased re-rupture rate on return to sport. The increased incidence of contralateral knee ACL injury in patients who have had BPTB reconstructions is reported in the literature28 and could be interpreted as an indication that this population is more likely to return to at-risk sports.


Quadriceps function

The cause of quadriceps weakness following ACL reconstruction is thought to be related to rehab deficiency rather than donor site. Studies have shown no difference in the quadriceps weakness between patients treated with allograft or autograft BPTB grafts.29 30 Similarly, studies have identified no difference in weakness when hamstrings grafts were compared with BPTB.20

Kneeling pain

The problem of kneeling pain is well described and accepted as a complication more associated with BPTB graft harvest than hamstring. Donor site tenderness can be reduced by the use of bone graft impacted into the bone defects and/or a two-incision approach.31 32

Anterior knee pain

The perception of simple anterior knee pain being exclusively related to BPTB harvest seems to have arisen from early retrospective studies without a comparison group.33 In 2001 Eriksson et al published a comparison between the two graft types and found no difference in the incidence and level of anterior knee pain except in the context of kneeling.34 In a systematic review of randomised control trials in 2004, Spindler et al showed that eight out of nine of the studies they included showed no difference in the incidence of anterior knee pain.33

The causes of anterior knee pain are numerous and include quadriceps inhibition and weakness, ACL deficiency and iatrogenic sensory disturbance through damage to the infrapatellar branch of the saphenous nerve. This nerve injury is a complication described for both the BPTB19 and the hamstring graft.35 36 For BPTB harvest the risk of damage to this nerve has been shown to be significantly reduced through a two-incision subcutaneous technique.32 The proximity of the nerve to the gracilis tendon may make avoidance of the nerve in hamstrings harvest more difficult and injury rates have been quoted at 55% in a recent paper,36 with only 20% experiencing improvement over the initial 6 months.


It has been suggested in the past that ACL reconstruction with BPTB grafts predisposes to arthritis. In a recent retrospective review, however, Liden et al found similar appearances in both hamstrings and BPTB grafts at 7 years.37 Similarly, 5 year results from Finland have not shown a difference between hamstring and BPTB grafts.38 This differs from the 10 year results seen in a recently published study by Pinczewski et al, where higher rates of arthritis were seen in BPTB grafts.39 This finding was also seen at 5 and 7 years.28 40

Logically, there is no obvious explanation for a BPTB graft being associated with a greater risk of arthritis than hamstrings. Both grafts are harvested in an extra-articular fashion, there is no demonstrable difference in quadriceps function and, when a difference in stability is found, the BPTB is usually identified as the more stable graft. It may, therefore, be that a different reason exists for the finding of increased osteoarthritis seen in the Pincewski series, and we would propose the following possible explanations:

  1. The BPTB grafts were fixed with interference screws, with the first three turns of the screw made with the knee flexed “until a solid hold had been made”; the knee was then extended. This may have resulted in a slightly captured knee and increased joint pressures. Whilst there was no difference in range of movement between the groups at 2 years, it is possible that increased contact pressure at full hyperextension had been an initial problem that triggered chondral degeneration that had yet to manifest at the 2 year stage.

  2. All of the cases in the Pincewski series were operated on within 12 weeks of injury. It is well known that the onset of arthritis is closely linked to the chondral status at time of reconstruction. It may be possible that chondral damage resulting from initial injury is not apparent at such an early time point.

In our opinion the use of arthritis in the decision-making process is unreliable. It is difficult to explain how a more secure and anatomical graft would predispose to arthritis. Any difference in arthritis is likely to be small and a large study would be required to accurately quantify any discrepancy.

In summary, we prefer the use of the BPTB graft as we feel it offers the optimum combination of graft integrity and maintenance of motor function. We accept that there are increased technical difficulties when compared with the hamstrings graft and also that the incidence of discomfort on kneeling is increased. We do not currently accept that BPTB graft use predisposes to arthritis when used appropriately. It is our view that the increased rate of laxity seen in the hamstrings group, along with demonstrable hamstring compromise and the high rate of tunnel widening, is sufficient reason to avoid the hamstrings graft except when absolutely necessary. In all but a few cases where a discomfort on kneeling would be a significant hindrance or in the skeletally immature patient we recommend BPTB as the graft of choice.


View Abstract


  • Competing interests: None.

Linked Articles

  • PostScript
    BMJ Publishing Group Ltd and British Association of Sport and Exercise Medicine