Bankart or Latarjet? Glenoid Bone Loss, ISIS and the Glenoid Track

A young patient has had a third anterior dislocation, and the question on the list is Bankart vs Latarjet: repair the labrum, or transfer the coracoid? The short answer is that bone decides. An arthroscopic Bankart repair is a soft-tissue operation and works when the glenoid arc is close to intact and the Hill-Sachs lesion stays on-track. Once glenoid loss reaches the critical range, or a deficient glenoid is paired with an engaging humeral lesion, a bony procedure such as the Latarjet is the operation that holds. This post sets out how to measure that, the two thresholds, the glenoid track, the ISIS, and what the trials show. The clinical background is on the anterior shoulder instability topic page.

Why does bone loss decide the operation?

A Bankart repair reattaches the anteroinferior labrum and capsule to the glenoid rim. It restores a soft-tissue bumper and ligament tension, but it cannot restore a rim that is no longer there. The paper that made this concrete is Burkhart and De Beer (2000): 194 consecutive arthroscopic Bankart repairs, 101 of them in contact athletes, mostly South African rugby players.

They defined significant bone defects as an inverted-pear glenoid or an engaging Hill-Sachs lesion. Recurrence was 4% in the 173 shoulders without them and 67% in the 21 with them. In contact athletes the figures were 6.5% and 89%. Their conclusion was that patients with significant bone deficits are not candidates for arthroscopic Bankart repair, and that with significant glenoid loss the surgeon should consider a Latarjet with a large coracoid graft.

Bone loss is also progressive. In the MOON Shoulder Instability cohort of 892 patients (Hettrich and colleagues, 2023 (opens in a new tab)), the number of prior dislocations was the factor most strongly associated with glenoid loss, Hill-Sachs lesions and combined lesions. Each additional dislocation raised the odds of a combined lesion by 95%. A shoulder that would have done well with a Bankart repair after the second episode may need a bony procedure after the sixth. The Bankart lesions page covers the soft-tissue lesion itself.

How is glenoid bone loss measured?

The view. The inferior glenoid is normally close to a circle. On an en-face (sagittal-oblique) 3D reconstruction, a best-fit circle is drawn on the intact inferior glenoid and the defect is expressed as a percentage of it. 3D CT is the reference standard; 3D MRI has been validated as an alternative and spares the radiation. In unilateral instability, the contralateral glenoid can serve as the template.

The method. The two common methods do not give the same number for the same defect. The width (linear) method expresses the defect width as a fraction of the circle's diameter, and tends to read slightly higher. The surface-area (Pico) method expresses the missing area as a fraction of the circle. Any percentage quoted against a threshold should state which was used.

Plain films. They do not quantify loss, but they screen for it. The West Point axillary and Bernageau views show the anteroinferior rim; the AP in external rotation and the Stryker notch view profile the Hill-Sachs lesion. Two of the ISIS items, worth four of its ten points, come from the AP radiograph.

At arthroscopy. The inverted pear corresponds to substantial loss, conventionally quoted at about 25–30%. The full measurement technique, with worked CT figures, is on the glenoid bone loss assessment page.

Critical and subcritical: what do 20% and 13.5% mean?

The classical critical threshold is about 20–25% glenoid loss. Above it, an isolated soft-tissue Bankart repair fails too often, and bony reconstruction is the standard answer. Di Giacomo, Itoi and Burkhart described glenoid grafting at 25% or more as the consensus of recent authors.

The subcritical figure comes from Shaha and colleagues (2015): 72 patients (73 shoulders) after isolated arthroscopic Bankart repair at a single military institution, divided into quartiles of glenoid loss, followed for a mean of 48 months. Re-dislocation did not differ across the lower three quartiles (7.3% combined) and rose to 27.8% in the top quartile (20–35.5% loss). So 20% is the right number if the question is whether the shoulder will dislocate again.

Function told a different story. WOSI scores worsened with every quartile, and above 13.5% loss they reached a level consistent with a poor clinical outcome. This held even in patients who never re-dislocated. A shoulder that stays in after a Bankart repair at 17% loss has not necessarily done well.

Two cautions keep this from becoming a rule. The cohort was young, military and under high mandatory activity, close to the worst case, and the study was a level III retrospective cohort. And the long view is mixed: in Lau and colleagues' 17-year follow-up of 75 patients after arthroscopic Bankart repair, 29% had re-dislocated, but redislocation was not related to the initial severity of glenoid loss. Function was impaired with loss of 7% or more, which the authors judged not severe enough to warrant a different operation. The threshold debate is not settled, and the honest position is a band of increasing concern from about 13.5% upwards, weighed against the patient.

What is the glenoid track?

Glenoid loss is half of a bipolar problem. The other half is the Hill-Sachs lesion, the posterolateral humeral head impaction, and the two interact. A Hill-Sachs lesion causes instability only if the glenoid rim can drop into it.

Yamamoto and colleagues (2007) (opens in a new tab) mapped glenoid contact on the humeral head in nine cadaveric shoulders in abduction, external rotation and horizontal extension. The contact zone, the glenoid track, had a medial margin about 18.4 mm from the cuff footprint, equivalent to 84% of the glenoid width. A Hill-Sachs lesion that extends medial to that margin can engage. Di Giacomo, Itoi and Burkhart (2014) turned this into the on-track/off-track method used clinically:

• Glenoid track width = 0.83 × the diameter of the inferior glenoid circle, minus the width of the anterior glenoid defect.
• Hill-Sachs interval (HSI) = the width of the Hill-Sachs lesion plus the bone bridge between its medial edge and the cuff footprint.
• If the HSI is greater than the track, the lesion is off-track and engages. If it is smaller, it is on-track.

Because the glenoid defect is subtracted from the track, the same Hill-Sachs lesion can move from on-track to off-track as the glenoid erodes. That is the bipolar interaction in a single calculation. The humeral side is covered in Hill-Sachs lesions.

Working the figure. In panel A the circle diameter is about 29 mm and the defect about 9 mm, so the track is 0.83 × 29 − 9, roughly 15 mm. The HSI in panel B is about 30 mm, twice the track: off-track. In panel C the diameter is about 24.5 mm with no defect, giving a track of about 20 mm, and the HSI in panel D is about 16 mm: on-track.

Does it predict failure? At Balgrist, Schwihla and colleagues (2023) applied the method to 163 shoulders after isolated arthroscopic Bankart repair, followed for a mean of 124 months. Recurrent instability (dislocation, subluxation or apprehension, or revision) occurred in 74% of off-track shoulders and 27% of on-track ones. Revision for instability was 48% against 13%. Their conclusion: an off-track lesion should be converted to on-track at surgery.

What does the ISIS add?

Bone measurements say nothing about the patient's age or sport. Balg and Boileau's Instability Severity Index Score combines both into a 10-point preoperative score, derived from 131 patients after arthroscopic Bankart repair with a mean follow-up of 31 months.

A score over 6 carried a recurrence risk of 70% after arthroscopic Bankart repair, and the authors stopped offering it to those patients in favour of a Bristow-Latarjet. The score was tested retrospectively on the same population that generated it, and external validation has been variable. In the Finnish randomised trial below, mean ISIS was under 3 in both arms, yet 21% of the Bankart group re-dislocated within two years, and neither the ISIS nor its components was significantly associated with early failure. Use the ISIS to flag the high-risk patient; use the CT to decide what the bone needs.

Other modifiers sit outside the score: a first dislocation before 20, many prior dislocations, seizures (which drive very high recurrence and bilateral disease), generalised laxity, and occupational demand in soldiers and collision athletes.

What does the trial evidence show?

A randomised trial of primary surgery. Kukkonen and colleagues (2022) randomised 122 men aged 16–25 with recurrent traumatic anteroinferior instability after a first dislocation treated non-operatively, across eight Finnish public hospitals. Bony Bankart and Hill-Sachs lesions were not exclusions; about a third of each arm had a glenoid defect the trial defined as significant (defect length at least half the glenoid width on en-face CT). At two years, with 91 patients analysed (25% loss to follow-up):

• Redislocation: 21% (10 patients) after arthroscopic Bankart against 2% (1) after open Latarjet (p=0.006).
• Return to the previous top level of competitive sport: 1 patient (9%) against 5 (56%).
• No significant difference in any other secondary outcome, including WOSI.

The population was young men only, follow-up was two years, and the dropout rate was high, so the result does not transfer directly to older or lower-demand patients.

Meta-analyses. An earlier review (An and colleagues, 2016 (opens in a new tab)) pooled eight retrospective comparative studies and found lower recurrence after Latarjet. A 2024 meta-analysis by Hossein Zadeh and colleagues included 21 studies (3 RCTs) and 13,176 shoulders. After arthroscopic Bankart the risk of recurrence and revision for instability was about three times higher (RR 3.08, 95% CI 2.03–4.68). The Bankart group had about 47% fewer total complications (RR 0.53, 95% CI 0.31–0.90).

What the Latarjet costs. In Griesser and colleagues' systematic review of 45 level IV studies (1,904 shoulders), the total complication rate after Bristow-Latarjet was 30%. Recurrent dislocation was 2.9%, recurrent subluxation 5.8%, and unplanned reoperation about 7%. The musculocutaneous and axillary nerves are most at risk, and the suprascapular nerve is endangered by screws exiting posteriorly. Graft position drives the late problems: a graft placed too lateral risks arthritis, one placed too medial risks recurrence.

Durability. Hovelius and colleagues (opens in a new tab) followed 118 Bristow-Latarjet repairs for 15 years: 98% of patients were satisfied and redislocation was rare. Pending their radiological follow-up, the authors recommended the operation only for revision after failed surgery, by surgeons familiar with it, so the series supports durability rather than routine primary use. The steps are in the Latarjet coracoid transfer operative guide.

Where does remplissage fit?

Remplissage fills the Hill-Sachs defect with posterior capsule and infraspinatus, converting an off-track lesion into an extra-articular, functionally on-track one. It is added to an arthroscopic Bankart repair, and it is the main alternative to the Latarjet when the humeral side is the dominant problem and glenoid loss is subcritical. Technique is on the remplissage page.

• Bankart with or without remplissage. In MacDonald and colleagues' RCT (108 patients, Hill-Sachs lesion with glenoid loss under 15%), recurrent instability at two years was 18% without remplissage against 4% with it, with 6 revisions against none. Patient-reported outcomes did not differ.
• Bankart plus remplissage against Latarjet. Haroun and colleagues (2020) pooled four comparative studies (379 patients, mostly level III) with engaging lesions and subcritical glenoid loss. Recurrence was similar (RR 0.72, 95% CI 0.37–1.41), and complications were about seven times more likely after Latarjet (RR 7.37). A later single-centre retrospective series of 108 patients with 20% loss or less reached the same conclusion in both primary and revision cases (Charles and colleagues, 2024 (opens in a new tab)).
• The trial that will answer it. STABLE is a multicentre RCT randomising 114 patients with recurrent instability and 10–20% glenoid loss (3D CT, best-fit circle) to Bankart plus remplissage or Latarjet, with WOSI at 24 months as the primary outcome. Its results are not yet published.

How do the numbers translate into an operation?

The rows overlap on purpose. Subcritical loss with an off-track lesion sits in two rows, and the choice between them rests on age, sport, laxity, the ISIS and the patient's view of a higher complication rate against a lower recurrence rate. For revision after a failed stabilisation, a systematic review of 16 studies (713 shoulders) found recurrence of 8.4% after open Latarjet, with redislocation in 0.9% (Ali and colleagues, 2021 (opens in a new tab)).

In the exam

Examiners rarely want a single answer. They want the reasoning in order: measure the bone, place the Hill-Sachs on the track, score the patient, then choose, and say what the alternative costs. Examination signs worth knowing for the clinical part are covered in shoulder special tests and their accuracy, and structuring the answer in passing the orthopaedic viva.