A retrospective review of patients over 18 years old with fixed severe thoracic hyperkyphosis (a deformity angle of over 70°) with a thoraco-lumbar apex (between T11 and L1) treated by simultaneous two-level thoracic PSO and thoraco-lumbar posterior fusion was performed. Follow-up evaluations were performed post-operatively, at 12 months and up to the final follow-up at 28.7 ± 2.5 months.

Informed consents for participation in the study and for the publication of clinical images were obtained from each patient.

Included patients were all affected by severe fixed thoracic hyperkyphosis with a thoraco-lumbar apex (between T11 and L1). Anterior bony fusion at the apex of the deformity was present in each case and documented by pre-operative CT scan. Fulcrum supine radiographs were taken in all cases to confirm the irreducibility of the deformity.

Patient demographics, the aetiology of the deformity and data on prior surgical treatments were collected by reviewing the medical records.

Operative time, blood loss, length of stay, and intra- and post-operative complications were recorded.

The deformity angle (used to evaluate the local spinal kyphosis angle, defined as the Cobb angle from the upper endplate of the proximal junctional normal vertebra to the lower endplate of the distal junctional normal vertebra) and the T1–T12 thoracic kyphosis (TK) and L1–S1 lumbar lordosis (LL) angles were measured on pre- and post-operative full-length standing radiographs. The C7 plumbline (C7PL)/central sacral vertical line (CSVL) and sagittal vertical axis (SVA) were used to assess coronal and sagittal imbalance. The pelvic incidence (PI), pelvic tilt (PT), sacral slope (SS), PI–LL mismatch (PI-LL) and GAP score [5] were used to evaluate the spinopelvic balance.

The sagittal deformity angular ratio (S-DAR) [6] was used pre-operatively to evaluate the radius of curvature of the deformity, and was calculated by dividing the deformity angle by the number of vertebral bodies in the deformed area.

Osteotomy healing was assessed in all cases with a CT scan performed at the 1-year follow-up evaluation.

The Oswestry Disability Index (ODI) was administered pre-operatively and at last follow-up.

Surgery was planned based on pre-operative full standing antero-posterior and lateral X-rays of the entire column and on CT scan; these guided the decision regarding the location of the osteotomies and the desired corrective angle, the fusion area, and eventually the need to obtain patient-specific guides for screw insertion.

The first osteotomy was always planned at the apex of the deformity, while the second was always planned at three levels above in order to maintain at least three pairs of screws between the two osteotomies. The degree of correction for each osteotomy was accurately measured on the 3D CT scan reconstruction.

Regarding the fusion area, the lower instrumented vertebra (LIV) was the sagittal stable vertebra according to the definition by Cho et al. [7]. The fusion was eventually extended distally in the presence of documented disc degeneration at the level below the planned LIV (Fig. 1). The upper instrumented vertebra was T3 in all cases.

A 54-year-old female (case 1) with severe iatrogenic thoraco-lumbar hyperkyphosis (prior uninstrumented anterior and posterior thoraco-lumbar fusion): pre-operative radiographic and clinical presentation

In the presence of an altered posterior anatomy (related to previous surgeries) with the presence of posterior spinal fusion mass, customized guides (Myspine, Medacta International SA, Castel San Pietro, Switzerland) for the placement pedicle screws were developed (Fig. 2); otherwise, the freehand technique was used.

Patient-specific guides were used for pedicle screw placement because of the abnormal anatomy of posterior elements due to prior fusion

The patient was placed prone on an Allen table; neurophysiologic monitoring with somatosensory-evoked potentials and trans-cranial motor-evoked potentials was used in all cases. After a standard posterior midline approach with longitudinal skin incision, careful subperiosteal exposure of the spinous processes and laminae was performed; when customized guides were used (in two cases), care was taken not to remove bone to avoid altering the contact points of the guides. Each guide was then placed on the corresponding vertebra and firmly held; then, after checking the contact surfaces, lateral and contralateral awls were used to prepare the entry point, and a 2.5 mm drill was used to prepare the pathway for the screws in the pedicle [8]. Finally, after tapping, polyaxial screws of an appropriate length and diameter were inserted according to the planned model. When customized guides were not needed (in one case), a freehand technique was used [9].

Screws were placed at all levels, except for those in planned osteotomies. The screw position was checked with a fluoroscope.

Then, the first thoracic PSO was performed as planned, at the apex of the deformity (T12 in two cases, T11 in one case). Laminectomy was performed at the level of the osteotomy and at the adjacent levels, one above and one below, to prevent buckling of the spinal cord during the correction manoeuvre; the pedicles were visualized and drilled to maintain orientation. The ribs at the level of the osteotomy were carefully exposed (3–4 cm) and removed, disarticulating the costotransverse joints bilaterally (Fig. 3). The bone was all saved for later fusion. Then, careful dissection of the lateral vertebral body wall was carried out bilaterally.

Wide laminectomy, temporary rod placement and rib resection at T12

After positioning dural retractors to optimize the visualization of the vertebral body, two osteotomes were positioned superior and inferior with respect to the pedicle, and the desired osteotomy angle was checked under fluoroscope guidance (Fig. 4). A temporary rod was placed on the opposite side to the first surgeon to prevent sudden collapsing of the spine and possible translation while completing the osteotomy. Then, lateral vertebral body wall cuts were made with straight osteotomes in a precise wedge according to the desired degree of closure.

Surgical and radiographic pictures of the first PSO, performed at T12 using a bone scalpel

The apex of the wedge was at the cortical anterior vertebral body wall, which was carefully violated with a sharp osteotome to obtain the desired corrective angle. The bony wedge was then removed. The same procedure was performed on the opposite side. Thoracic roots were preserved in all cases. All soft tissues and osseous elements were carefully removed from the anterior portion of the dura to prevent any compression of the neural elements anteriorly during osteotomy closure (Fig. 5).

Careful assessment of the complete posterior wall resection in order to avoid dural sac entrapment during the subsequent osteotomy closure

PSO closure was then performed by gentle compression across temporary rods and by adjusting the Allen bed frame to reduce the kyphosis.

The same procedure was repeated on the other thoracic vertebra (T8 in two cases, T7 in one case) to achieve proper correction of the deformity (Figs. 6, 7).

The second PSO is performed at T8

Careful assessment of the complete posterior wall resection and subsequent osteotomy closure

Then, the temporary rods on the right side were removed and gently replaced with a pre-bent 5.5 mm titanium rod; the same procedure was performed on the left side. After placing the definitive rods (Fig. 8), segmentary compression was applied to the screws and rods to achieve bone-to-bone closure.

Definitive rods are placed after both osteotomies are closed

Careful decortication of the posterior spinal elements of the instrumented vertebrae was followed by apposition of allograft bone. A subfascial (Fig. 9) drain was placed and a standard suture was performed.

Case 1: postoperative radiographic and clinical presentation

Early mobilization out of bed started on post-operative day 1. For the first 12 weeks after surgery, a thoracic lumbar sacral orthosis (TLSO) to restrict spinal movements and facilitate initial bone graft fusion was prescribed.

Three patients (two females and one male) were included. The average age was 42.6 (range 30–52) years. Patient characteristics are summarized in Table 1. According to PI, the Roussouly morphotype of all patients was 4 [10].

Each variable was presented as mean ± SD (standard deviation). Statistical analyses were performed using paired t-tests (SPSS version 17.0). Normality was assumed, and a P value < 0.05 was considered significant.