Fifty patients were treated with VEPTR™ constructs for thoracic insufficiency syndrome at our center between 2001 and 2007. Fourteen of these 50 patients had placement of a spine-to-spine construct using VEPTR™ implant in combination with standard spinal implants using a modification of the method described by Akbarnia et al. [1
] (Table ). The indications for the spine-to-spine constructs using dual rods were (1) absence of a primary chest wall deformity, (2) progression of spinal deformity to a Cobb angle of greater than 50°, and (3) migration of a previously placed proximal rib anchor or of a prior non-VEPTR™ growing rod to the point of loss of stable fixation. The contraindications were (1) severe proximal kyphosis preventing successful control of the proximal deformity and (2) lack of sufficient soft tissue coverage to allow for safe placement of a proximal device as judged by our team’s general and thoracic surgeon. Ten patients were female and four were male. One patient had prior growing implants placed at another center with two distal rod breakages and distal hook migration, and four had a VEPTR™ rib-to-spine construct with an average of two rib anchor failures before conversion to a VEPTR™ spine-to-spine construct. All patients classified as “anchor failures” were symptomatic with pain at the anchor failure site, clinically noticeable loss of correction of their deformity, and/or skin at risk due to implant prominence. Average age at primary implantation of a spine-to-spine construct was 6.4 years (range, 2.1–10.8 years). Minimum followup was 24 months (mean, 35 months; range, 24–48 months). All patients have been continuously followed without loss of followup. No patients were recalled specifically for this study; all data were obtained from medical records and radiographs. Data were collected under a prospective Institutional Review Board (IRB)-approved protocol.
Clinical data on 14 patients treated by spine-to-spine constructs
All surgery was performed by a single surgeon (KMS). The specific surgical technique used in these cases included removal of pre-existing implants when warranted. Subperiosteal exposure of the top and bottom two vertebrae was achieved for implant placement and fusion (Fig. ). Proximal and distal anchor levels typically included measured end vertebrae extending distally to one or two levels above the measured end vertebra. A superior anchor was established using a claw technique with transverse process and pedicle hooks from the Dual-Opening Universal Spine System (Synthes North America). Caudally, a two-level pedicle screw anchor was created. A ventriculoperitoneal shunt passer was used to create a subfascial opening between the incisions through which a 20-Fr chest tube was passed. The assembled implants were then advanced subfascially with the female rod end left long by approximately 2 cm at the caudal pedicle screw anchor. The overlapping male and female components were maximally overlapped for future lengthening potential. Implants were engaged proximally and distally, and an initial lengthening of 1 to 2 cm was achieved by distraction at the caudal anchor. Decortication and placement of allograft were performed at the anchor sites. Drains were not placed. The patient was molded for a custom thoracolumbosacral orthosis that was applied 2 to 3 days later and worn full time for 3 months. The first lengthening was performed at 3 months with subsequent lengthenings at scheduled 6-month intervals. Initial implantation was always performed using comprehensive neuromonitoring (transcranial motor-evoked potentials, somatosensory evoked potentials, and free-run EMGs). No monitoring was used for lengthenings or exchanges unless there were changes at the initial implantation or the child had a history of a neural axis abnormality such as a tethered spinal cord. Average implantation time was 2.5 hours (range, 1 hour 20 minutes to 3 hours) and average blood loss was 75 mL (range, 25–150 mL). No patient received a transfusion. Patients were discharged home once they were tolerating oral intake and only taking oral pain medications.
Limited exposure is required at the upper and lower ends of the spinal construct. Implants are passed subfascially through a space created by a shunt passer and chest tube.
An underarm custom-molded thoracolumbosacral orthosis was applied the second postoperative day and continued for 3 months whenever the child was in an upright position. No physical therapy was recommended except for the initial mobilization after surgery. All children were supervised by their parents in returning to their preoperative mobility status.
Patients had a wound check by a nurse at 1 to 2 weeks postoperatively and returned for an initial radiograph in a brace at 4 to 6 weeks after surgery. Initial expansion was planned for and performed at 12 weeks after initial implantation. Bracing was discontinued just before the first expansion.
Information related to patient age, diagnosis, preceding treatment, and ambulatory status before and after treatment was collected (Table ). Length of surgery, intraoperative blood loss, neuromonitoring changes, and postoperative complications including infections, implant failures, unplanned admissions to this or other hospitals, and unplanned returns to the operating room were recorded.
Three of us (KKW, BKD, KMS) independently measured the following variables on the preoperative and last followup radiographs: maximum coronal and sagittal Cobb angles; coronal T1–S1 length (superior end plate of T1 to superior end plate of S1); coronal balance (horizontal distance in millimeters from the center of T1 to the center sacral vertebral line); sagittal balance (distance in millimeters from a vertical line from the center of the superior end plate of S1 to the midpoint of the superior end plate of C7); and space available for the lung (SAL) as described by Campbell et al. [4
] (ratio of length of the line from the midportion of the first rib of the concave hemithorax to the midportion of the dome of the ipsilateral diaphragm to the length of the line from the midportion of the first rib of the convex hemithorax to the midportion of the dome of the ipsilateral diaphragm). We used the average from the three independent observers for each variable (Table ). An analysis of interobserver variability for all variables was performed using the intraclass correlation coefficient, with all being greater than 0.92.
Summary of results for 14 patients treated with spine-to-spine constructs
Descriptive statistics were used for nominal data. Radiographic measurements were compared over the selected time points using Friedman’s test for analysis of repeated measurements of nonparametric, continuous data.