Internal hemipelvectomy: A single institution's learning curve and longitudinal experience.

Background And Objectives: Wide margin resection for pelvic tumors via internal hemipelvectomy is among the most technically challenging procedures in orthopedic oncology. As such, surgeon experience and technique invariably affect patient outcomes. The aim of this clinical study was to assess how an individual surgeon's experiences and advancements in technology and techniques in the treatment of internal hemipelvectomy have impacted patient outcomes at our institution.

Methods: This study retrospectively examined a single tertiary academic institution's consecutive longitudinal experience with internal hemipelvectomy for primary sarcoma or pelvic metastases over a 26-year period between the years 1994 and 2020. Outcomes were assessed using two separate techniques. The first stratified patients into cohorts based on the date of surgery with three distinct "eras" ("early," "middle," and "modern"), which reflect the implementation of new techniques, including three-dimensional (3D) computer navigation and cutting guide technology into our clinical practice. The second method of cohort selection grouped patients based on each surgeon's case experience with internal hemipelvectomy ("inexperienced," "developing," and "experienced"). Primary endpoints included margin status, complication profiles, and long-term oncologic outcomes. Whole group multivariate analysis was used to evaluate variables predicting blood loss, operative time, tumor-free survival, and mortality.

Results: A total of 72 patients who underwent internal hemipelvectomy were identified. Of these patients, 24 had surgery between 1994 and 2007 (early), 28 between 2007 and 2015 (middle), and 20 between 2016 and 2020 (modern). Twenty-eight patients had surgery while the surgeon was still inexperienced, 24 while developing, and 20 when experienced. Evaluation by era demonstrated that a greater proportion of patients were indicated for surgery for oligometastatic disease in the modern era (0% vs. 14.3% vs. 35%, p = 0.022). Fewer modern cases utilized freehand resection (100% vs. 75% vs. 55%, p = 0.012), while instead opting for more frequent utilization of computer navigation (0% vs. 25% vs. 20%, p = 0.012), and customized 3D-printed cutting guides (0% vs. 0% vs. 25%, p =  0.002). Similarly, there was a decline in the rate of massive blood loss observed (72.2% vs. 30.8% vs. 35%, p = 0.016), and interdisciplinary collaboration with a general surgeon for pelvic dissection became more common (4.2% vs. 32.1% vs. 85%, p < 0.001). Local recurrence was less prevalent in patients treated in middle and modern eras (50% vs. 15.4% vs. 25%, p =  0.045). When stratifying by case experience, surgeries performed by experienced surgeons were less frequently complicated by massive blood loss (66.7% vs. 40% vs. 20%, p = 0.007) and more often involved a general surgeon for pelvic dissection (17.9% vs. 37.5% vs. 65%, p = 0.004). Whole group multivariate analysis demonstrated that the use of patient-specific instrumentation (PSI) predicted lower intraoperative blood loss (p = 0.040). However, surgeon experience had no significant effect on operative time (p = 0.125), tumor-free survival (p = 0.501), or overall patient survival (p = 0.735).

Conclusion: While our institution continues to utilize neoadjuvant and adjuvant therapies following current guideline-based care, we have noticed changing trends from early to modern periods. With the advent of new technologies, we have seen a decline in freehand resections for hemipelvectomy procedures, and a transition to utilizing more 3D navigation and customized 3D cutting guides. Furthermore, we have employed the use of an interdisciplinary team approach more regularly for these complicated cases. Although our results do not demonstrate a significant change in perioperative outcomes over the years, our institution's willingness to treat more complex cases likely obscures the benefits of surgeon experience and recent technological advances for patient outcomes.