A 16-year-old girl presented with left buttock and leg pain for 3 months. There was no neurological dysfunction. Pelvic magnetic resonance imaging (MRI) revealed a lobulated mass (3.4×4.7×5 cm, roughly 362 cm³) with peripheral enhancement at the left sacrum mainly at the level of S1 and S2 (Fig. 1A). Fluoroscope-guided bone biopsy was performed preoperatively. Histological examination revealed chondroblastic osteosarcoma. Treatment plan was decided in a multidisciplinary tumor board consisting of a pediatric oncologist, neurosurgeon, orthopedic surgeon, and radiation oncologist. Neoadjuvant chemotherapy with cisplatin, adriamycin, and methotrexate was given preoperatively for 3 months. Follow-up pelvic MRI demonstrated a moderate reduction (32%) of the mass (4.4×3.2×4.2 cm, roughly 248 cm³) with a significant reduction of the peripheral extension (Fig. 1B). The extent of previously noted enhancement at the left iliacus muscle was also decreased. During the period of neoadjuvant chemotherapy, implant design and surgical planning were conducted.

The pelvic MRI was exported in a digital imaging and communications in medicine format to the 3D software (Mimics; Materialise, Leuven, Belgium). The whole process is illustrated in Fig. 2. A customized volume was generated as stereolithography format, which is a standard form of manufacturing. The process was performed in a stepwise manner. The sacrum was segmented by threshold using a function that includes only those pixels of the image with a value higher than or equal to the threshold value (segmentation). A 3D rendering of the sacrum was reconstructed with an automatic function (3D reconstruction). A customized implant was designed using other software (3-matic; Materialise). The sacral implant was designed as a patient-specific structure mostly identical to the anatomical structure. Several rapid prototype (RP) models were printed using a conventional 3D-printer for review and correction. With the final MRI and RP models, modification was made interactively. Then, the internal architecture was designed using other software (Magics; Materialise).

A metal printer (Arcam-EBM A1; ArcamAB, Mölndal, Sweden) system was used to print the implant. The material was Ti-6Al-4V extra-low interstitial medical grade powder. In this system, the focused electron beam is rastered over each successive layer of powder, which is gravity-fed from powder containers and raked into successive layers roughly 50 µm thick. The building component is lowered on the build table with the completion of each successive layer. Each newly raked powder layer is initially rastered by the beam with approximately 11 passes at a beam current of approximately 35 mA to preheat each layer to approximately 600℃. This layer preheat is normally followed by a 4 mA beam current. The melt scan is driven by a 3D computer-assisted design program and melts only selected layer areas, which add metal to the build. The printed implant was inspected extensively and modified several times. The final implant was obtained and delivered to the operating room (Fig. 3).

Informed consent was obtained from the patient for operation and publication. After general endotracheal anesthesia, a unilateral retroperitoneal approach was made from the left side. The goal of our anterior approach was dissection of the ventral portion of L5 and the sacrum from the visceral and vascular structures. Anterior sacroiliac ligament was released and the left S1, S2, S3, and S4 nerve roots were ligated and cut. To ensure the dissection margin, dry tapes were placed around the lesion site. The patient's position was subsequently changed to prone on a surgical frame. A midline vertical incision was made from L2 to the level of the coccyx, and the paraspinal muscles were dissected. After full exposure of the dorsal sacrum, we performed laminectomy and cut the proximal portion of the left S1, S2, S3, and S4 nerve roots. Left S1 and S2 nerve roots were involved in tumor. We needed a space that the implant could be inserted all at once. Li, et al.5 reported that unilateral sacral nerve roots resection could preserve normal residual function. Thus, resection of left sacral nerve roots was inevitable. We cut the posterior sacroiliac ligamentous structures, and then split the sacrum on its exact midline. En bloc resection of the left hemisacrum was achieved with cutting the distal portion of the left S1, S2, S3, and S4 nerve roots (Fig. 3). Pedicle screws were inserted bilaterally from L3 to the iliac. One screw was inserted on a 3D-printed construct at the level of S1. The sacral screws and lumbo-iliac screws were strongly attached with rods and domino connectors. At the midline, the implant was further fixed to the right sacrum with metallic cables (Atlas cable; Medtronic, Memphis, TN, USA) through the sacral pore and to the iliac bone (Fig. 4A).

The patient was advised for 2 weeks bed rest. We encouraged the patient to change her position often to prevent surgical wound sores. The patient was able to walk after 2 weeks. Neurogenic bladder or bowel dysfunction was absent. Two weeks after surgery, the foley catheter was removed. Five days later, the patient did not have any bladder and bowel symptoms. Therefore, we did not need additional examination for bladder and bowel functions. However, a left foot drop and neuropathic pain on left leg occurred postoperatively because of resection of the left S1 nerve root. We applied an ankle-stabilizing orthosis and performed rehabilitation. Gabapentin was administered in order to control neuropathic pain. Visual Analog Scale (VAS) socre of left leg was eight immediately after the surgery. VAS score was gradually reduced to three at a year after the surgery. The pediatric oncologist had performed three cycles of adjuvant chemotherapy up to 12 months after the surgery. Complications were not observed until a year after the surgery. In our patient, plain radiographs of the lumbosacral spine showing the anterior-posterior and lateral views at 12 months postoperatively demonstrate that instrumentation has been well maintained (Fig. 4B and C). Also, bone ingrowth into the titanium porous structure and bone fusion between medial side of right sacrum and high-density aspect of the implant were confirmed by computed tomography (Fig. 5).