In this series of adolescent idiopathic scoliosis double curves we found postoperative transverse plane pelvic rotation increase (TPPRI) in 7 of 17, with resolution in 5 of the 7. Two of the variables studied were significantly associated with TPPRI. They were the addition of un-instrumented sequential anterior thoracolumbar/lumbar discectomy and arthrodesis and increased preoperative tilt of the vertebra below the lower instrumented vertebra. The occurrence of TPPRI and whether or not it persisted did not affect clinical outcome.
We interpret postoperative TPPRI to be a decompensation caused by extension of the corrective thoracolumbar/lumbar rotational load into the lumbosacral hemicurve below. This is supported by our finding that postoperatively the rotation of the vertebra below the lower instrumented vertebra in the direction of the thoracolumbar/lumbar curve (counterclockwise) had not changed in the no-TPPRI group, whereas in the TPPRI group it had decreased significantly, from 6° to 2°. Compared to the no-TPPRI group, the TPPRI group had significantly less postoperative rotation, 7° versus 2°. Transverse plane pelvis rotation increase re-compensated for 5 of 7 patients. Although we were unable to document the site of recompensation, at least partially because of the small number of patients, we believe that it occurred between the lower instrumented vertebra and the pelvis. The lack of postoperative to follow-up change in the coronal curves supports this.
Coronal plane decompensation also occurred in the TPPRI group and later resolved. Coronal plane recompensation appeared to occur at the disc below the lower instrumented vertebra. From postoperative to follow-up the tilt of the lower instrumented vertebra decreased from -4° to -1° in the TPPRI group while the tilt of the vertebra below was unchanged at 7°. This left the TPPRI group with an average of 6° disc wedge below the lower instrumented vertebra, which was the same for all 17 patients as a group.
Thus, both the transverse and coronal plane decompensation likely compensated in the junctional region just below the lower instrumented vertebra. We believe that recompensation was possible because this region had not been included in the instrumentation and arthrodesis.
Decompensation following transmitted rotation loading was noted soon after the introduction of Cotrel-Dubousset instrumentation [1, 19]. Although the mechanism is the same as TPPRI, it was different as the rotation was transmitted from the thoracic spine, where it was corrective, to the upper lumbar spine, where it was deforming. Instrumenting this transitional zone both locked the deforming rotation in place and prevented recompensation through a mobile transition zone [1, 19]. TPPRI, on the other hand, was occurring below the instrumented spine, leaving the transitional zone free to recompensate .
Although TPPRI has not been reported as far as we can tell, Dubousset has emphasized the concept of the pelvic vertebra, with six degrees freedom of motion between the hips and lumbosacral joint . It is unlikely that TPPRI has gone unnoticed. However, it is difficult to document and quantify, does not appear to affect the patient's perceived quality of life, and usually resolves.
Double curve treatment results are seldom reported separately [15, 22]. An exception is the series reported by Yeon et al. . In 15 patients with Lenke 3 curves they tested the hypothesis that addition of anterior instrumentation to the anterior procedure would "more effectively correct and maintain normal coronal alignment in the distal unfused spine." The 7 patients treated with supplemental anterior instrumentation also had posterior instrumentation stopping at L3. In comparison, 6 of 8 treated without the addition of anterior instrumentation had posterior instrumentation to L4. Indeed, the tilt of L4 pre- and postoperative was the same for the 2 groups and was similar to ours without the addition of anterior instrumentation.
We believe TPPRI may have come to our attention at least partially because of our goals to never instrument below lumbar 3 or its equivalent and to leave the lower instrumented vertebra as normally aligned as possible. To accomplish this, we sought as complete correction of the thoracolumbar/lumbar curve as possible. Several measurements suggest that for the group this was pretty well accomplished: thoracolumbar/lumbar ATI improved from 16° to 1°, lower instrumented vertebra tilt from -26° to -3°, and the thoracolumbar/lumbar Cobb from 62° to 17°.
We realize the literature is not clear that instrumentation to L3, and the better instrumented curve correction necessary, is better than instrumentation to L4 in the long term. Our experience is the same as that reported by Islam et al. that the majority of previously operated scoliosis patients requiring surgical treatment of lower adjacent degeneration were originally instrumented and fused to L4 . However, the L3-L4 motion segment is often at the junctional zone between curves, thus more mobile and less stable. Dubousset has recommended against stopping instrumentation and arthrodesis above the more mobile motion segment . It will probably be a long time before the tradeoff between greater 3-planar correction and stopping at L3 or whether the less correction necessary with stopping at L4 is known.
We were disappointed that we could not develop more specific guidelines for adding supplemental anterior thoracolumbar/lumbar discectomy and arthrodesis in order to gain better correction. Patients with thoracolumbar ATI of ≤ 11° are less likely and those of ≥ 17° more likely to benefit from the supplemental surgery. The same can be said of those with a LIV +1 to sacrum Cobb angles of ≤ 18° and ≥ 23°. A possible relative variable that we could not quantify or evaluate is the "art" factor. The instrumentation sequence is complicated, and it is possible that it is not applied with equal effectiveness, even by the same surgeon in the same surgical environment.
It has been suggested that this TPPRI phenomenon is unique to the brand of instrumentation used. We have no material to make a direct comparison. However, we believe the finding is related to the effectiveness of direct spine rotation and that the type of instrumentation used to achieve it is immaterial.
Our study is open to several criticisms. Our method of transverse plane pelvis rotation measurement is not precise. Anticipatory CT imaging is unlikely for a new observation and unjustified given the apparent lack of clinical importance of TPPRI. Our method does allow useful quantification to be made from clinically available radiographs. As documented in this and our previous study, the intra-observer, inter-observer, and positioning reliabilities are good. Our method also cannot compensate for intrinsic pelvic asymmetry. For 2 reasons these shortcomings do not appear to be detrimental to this study. First, the baseline preoperative measurement was nearly neutral. Second, the L/R hemipelvis ratios were also compared for change. And, constrained patient positioning, especially if it included the pelvis, would possibly/probably mask this largely temporary transverse plane pelvic decompensation. Evolving, clinically practical three-dimensional imaging technology will make it possible to address these criticisms in future series.
These findings have to be considered preliminary because of the small number of patients operated. The observed marginal significance in the preoperative TL/L ATI's when comparing the posterior only with the anterior then posterior surgery groups in part may be due to the small number of patients in the study. To confirm the observed difference of the preoperative TL/L ATI's (11 ± 4.1° and 17 ± 4.8°) between the 2 groups as being significant at an alpha level of 0.05 (2-sided) and a statistical power of 80%, one must have at least 10 patients per group. To account for multiple comparisons (or an alpha of 0.01; 2-sided), one must have at least 15 patients per group at the same level of an 80% statistical power.