Objective:

In double and triple major adolescent idiopathic scoliosis curves it is still controversial whether the lowest instrumented vertebra (LIV) should be L3 or L4. Too short a fusion can impede postoperative distal curve compensation and promote adding on (AON). Longer fusions lower the chance of compensation by alignment changes of the lumbosacral curve (LSC). This study sought to improve prediction accuracy for AON and surgical outcomes in Lenke type 3, 4, and 6 curves.

Methods:

This was a retrospective multicenter analysis of patients with adolescent idiopathic scoliosis who had Lenke 3, 4, and 6 curves and ≥ 1 year of follow-up after posterior correction. Resolution of the LSC was studied by changes of LIV tilt, L3 tilt, and L4 tilt, with the variables resembling surrogate measures for the LSC. AON was defined as a disc angle below LIV > 5° at follow-up. A matched-pairs analysis was done of differences between LIV at L3 and at L4. A multivariate prediction analysis evaluated the AON risk in patients with LIV at L3. Clinical outcomes were assessed by the Scoliosis Research Society 22-item questionnaire (SRS-22).

Results:

The sample comprised 101 patients (average age 16 years). The LIV was L3 in 54%, and it was L4 in 39%. At follow-up, 87% of patients showed shoulder balance, 86% had trunk balance, and 64% had a lumbar curve (LC) ≤ 20°. With an LC ≤ 20° (p = 0.01), SRS-22 scores were better and AON was less common (26% vs 59%, p = 0.001). Distal extension of the fusion (e.g., LIV at L4) did not have a significant influence on achieving an LSC < 20°; however, higher screw density allowed better LC correction and resulted in better spontaneous LSC correction. AON occurred in 34% of patients, or 40% if the LIV was L3. Patients with AON had a larger residual LSC, worse LC correction, and worse thoracic curve (TC) correction. A total of 44 patients could be included in the matched-pairs analysis. LC correction and TC correction were comparable, but AON was 50% for LIV at L3 and 18% for LIV at L4. Patients without AON had a significantly better LC correction and TC correction (p < 0.01). For patients with LIV at L3, a significant prediction model for AON was established including variables addressed by surgeons: postoperative LC and TC (negative predictive value 78%, positive predictive value 79%, sensitivity 79%, specificity 81%).

Conclusions:

An analysis of 101 patients with Lenke 3, 4, and 6 curves showed that TC and LC correction had significant influence on LSC resolution and the risk for AON. Improving LC correction and achieving an LC < 20° offers the potential to lower the risk for AON, particularly in patients with LIV at L3.