Risk factors for postoperative pathological fractures in lower extremity soft tissue sarcoma patients undergoing limb-sparing surgery

Abstract

BACKGROUND

Soft tissue sarcomas (STSs) of the lower extremities are rare malignancies increasingly treated with limb-sparing surgery combined with radiotherapy (RT). While this approach provides excellent oncologic control and preserves function, it is associated with late skeletal complications. Postoperative pathologic fractures (PPFs) represent a serious and potentially disabling outcome, often requiring additional surgical intervention and impairing quality of life. Reported fracture rates and associated risk factors vary widely, and contemporary data incorporating modern surgical and radiotherapeutic techniques remain limited.

AIM

To identify independent predictors of PPFs following limb-sparing surgery for lower extremity STS.

METHODS

This retrospective cohort study included 266 patients with intermediate- or high-grade lower extremity STS treated with limb-sparing surgery between 2010 and 2020. Clinical, tumor-related, surgical, and treatment variables were collected. PPF was defined as a low-energy fracture occurring within the surgical field. Cumulative fracture incidence was estimated using Kaplan-Meier analysis. Risk factors were assessed using univariable analyses and multivariable Cox regression with Firth’s penalized likelihood.

RESULTS

During a median follow-up of 48.9 months, 23 postoperative pathological fractures occurred. The cumulative fracture incidence was 1.8% at 1 year, 13.0% at 5 years, and 20.6% at 10 years. Fractures most frequently involved the femur. In univariable analyses, older age, arterial hypertension, larger tumor size, RT, and bone-involving surgical procedures were associated with increased fracture risk. In multivariable analysis, bone resection and osteoporosis remained independently associated with PPF, whereas RT, age, and arterial hypertension were not statistically significant. Tumor size was not included in the multivariable model due to lack of significance in univariable analysis but remains clinically relevant.

CONCLUSION

PPFs are a relevant late complication following limb-sparing surgery for STS. Fracture risk is primarily associated with bone-involving surgical procedures and osteoporosis, whereas RT shows an elevated but not statistically significant association.

Key Words: Soft tissue sarcoma; Limb-sparing surgery; Postoperative pathological fracture; Radiotherapy; Bone resection

Core Tip: Postoperative pathologic fractures are an underrecognized late complication following limb-sparing surgery for lower extremity soft tissue sarcoma (STS). In this retrospective cohort of 266 patients with long-term follow-up, fractures occurred predominantly in the femur and often several years after treatment. Partial bone resection and osteoporosis were independently associated with increased fracture risk, whereas radiotherapy showed an elevated but not statistically significant association. These findings underscore the importance of long-term surveillance and careful risk stratification in patients undergoing limb-sparing management for STS.

INTRODUCTION

Soft tissue sarcomas (STSs) are rare malignant tumors of mesenchymal origin, accounting for less than 1% of adult cancers worldwide[1]. The lower extremities are the most common site, and treatment typically consists of wide local excision with adjuvant radiotherapy (RT), occasionally combined with chemotherapy for locally advanced or high-grade disease[2]. Over recent decades, limb-sparing surgery has largely replaced amputation in most cases, providing comparable oncologic control with superior functional and psychosocial outcomes[3].

While limb preservation represents a major therapeutic advance, it is associated with distinct long-term musculoskeletal complications. Among these, postoperative pathologic fractures (PPFs) are particularly debilitating, often occurring months or years after treatment in previously irradiated and/or surgically compromised bone. These fractures can lead to significant pain, impaired mobility, and the need for additional surgical interventions, thereby compromising quality of life and functional independence[4-6].

Several potential risk factors for PPF have been proposed. RT is known to impair bone vascularization and remodeling, leading to dose-dependent reductions in bone strength and elasticity[7-10]. Surgical factors, such as periosteal stripping or cortical bone resection, directly compromise biomechanical stability and load distribution[11]. Tumor-related characteristics, including size and histology, may necessitate more extensive resections and higher radiation doses, thereby amplifying cumulative skeletal injury[12]. Patient-related factors, such as osteoporosis, metabolic bone disease, and advanced age, may also contribute[13].

However, existing evidence remains limited. Most prior studies are characterized by small sample sizes, heterogeneous patient populations, or incomplete adjustment for confounding variables in multivariate analyses[14-16]. Consequently, reported fracture incidences vary widely, ranging from 2% to 25%, and the relative contribution of individual risk factors remains unclear[17]. Moreover, advances in RT techniques, such as intensity-modulated RT (IMRT), and refinements in surgical strategies underscore the need for contemporary, methodologically robust data.

Therefore, this study aimed to identify independent predictors of PPF in a large, systematically analyzed cohort of patients with lower extremity STSs undergoing limb-sparing surgery at a single tertiary sarcoma referral center. In doing so, we sought to improve risk stratification, inform surgical and radiotherapeutic planning, and guide postoperative surveillance and patient counseling.

MATERIALS AND METHODS

Study design and patient selection

This retrospective cohort study included all patients presenting with suspected STS to the outpatient clinic of the Department of Orthopedics between January 2010 and December 2020. During this period, 1443 patients were screened for eligibility, of whom 266 met the inclusion criteria and were included in the final analysis. Eligible patients underwent limb-sparing surgery for histologically confirmed malignant intermediate- or high-grade (G2/G3) STS of the lower extremity, including tumors involving the pelvic bones, at our institution. Patients with low-grade sarcomas, benign tumors, non-limb-sparing procedures, or incomplete surgical management at our center were excluded.

Data collection and follow-up

Clinical data were extracted from the institutional electronic patient database, including pathology reports, operative notes, and follow-up records. Missing data were supplemented with archived medical correspondence, imaging reports, and external medical documentation when available. Follow-up duration was calculated from the date of index surgery to the date of the last clinical or radiographic assessment.

Definition and classification of PPF

PPF was defined as a fracture occurring within the anatomical region of the prior surgical field and associated with minimal or no trauma. Femoral fracture locations were classified based on postoperative radiographic imaging.

Variables and covariates

Patient-related variables included age, sex, height, weight, and body mass index (BMI). Pre-existing comorbidities included osteoporosis, rheumatoid arthritis, neurofibromatosis type 1, diabetes mellitus, renal insufficiency, and arterial hypertension. Treatment-related variables included receipt, timing, modality, total dose, number of fractions, and dose per fraction of RT, as well as administration of chemotherapy. To ensure biological comparability across varying radiation schedules, all doses were converted to the equivalent dose in 2 Gy fractions, assuming an α/β ratio of 3 Gy for bone tissue.

Tumor-related variables comprised histological subtype, anatomical location, maximum tumor diameter, histological grade, and resection margin status (R0, R1, R2, or Rx). Surgical variables included periosteal stripping, partial bone (tangential) resection, and tumor location in relation to the fascia (epifascial vs subfascial). Periosteal stripping was defined as partial, tumor-dependent deperiostization of the affected bone surface, typically performed tangentially without removal of osseous tissue. Partial bone (tangential) resection was defined as removal of a superficial cortical layer (cortical shaving) without creation of a cortical window and without full-thickness cortical resection, thereby preserving the structural continuity of the remaining cortex.

Statistical analysis

Time-to-event analyses were performed to account for the longitudinal nature of PPFs. The cumulative incidence of PPF was estimated using the Kaplan-Meier method. Time-to-event was defined as the interval between the date of surgery and the occurrence of the first PPF, or the date of last follow-up for patients without fracture. Differences between groups were assessed using the log-rank test. Kaplan-Meier analyses included 248 patients, with 18 patients having follow-up shorter than 1 month being excluded to meet the requirements of time-to-event analysis.

Univariable associations were evaluated using log-rank tests and Cox proportional hazards regression, with results reported as hazard ratios (HRs) and 95%CIs. STS histological subtype was compared using Fisher’s exact test. The predictive performance of tumor size for fracture risk was assessed using receiver operating characteristic (ROC) curve analysis, and the optimal cutoff value was determined using the Youden index.

To account for the low number of events and reduce the risk of model overfitting, multivariable analysis was performed using Cox regression with Firth’s penalized likelihood. Variables were selected based on clinical relevance and the results of univariable time-to-event analyses.

Statistical analyses were performed using R (version 4.4.2; R Core Team, 2024), with the following packages: Survival (version 3.6-4), survminer (version 0.5.1), pROC (version 1.19.0.1), dplyr (version 1.1.4), janitor (version 2.2.1), readxl (version 1.4.3), writexl (version 1.5.4), gtsummary (version 1.7.2), rempsyc (version 0.1.7), plyr (version 1.8.9), and ApaTables (version 2.0.8). A two-sided P value < 0.05 was considered statistically significant.

RESULTS

Patient characteristics and follow-up

The median follow-up was 48.9 months (range: 0-162 months). Of the 266 patients, 150 (56.4%) were male and 116 (43.6%) were female. The median age at surgery was 60 years [interquartile range (IQR): 45-73 years], including 4 patients younger than 18 years. Median BMI, available for 182 patients, was 26.8 (IQR: 24.6-30.2).

Sixty-seven patients had undergone prior non-curative surgery (Rx, R1, or R2) elsewhere, and 18 had been diagnosed with STS at least 1 year before referral. Smoking status was documented for 127 patients, of whom 10 (7.9%) were current smokers and 27 (21.3%) former smokers.

At the time of surgery, arterial hypertension was present in 110 patients (41.6%), renal insufficiency in 31 (11.7%), diabetes mellitus in 30 (11.3%), and osteoporosis in 5 (1.9%). Neurofibromatosis type 1 was documented in 3 patients (1.3%), and rheumatoid arthritis in 2 (0.7%).

Tumor characteristics

Tumors were most commonly located in the thigh (n = 173), followed by the lower leg (n = 45), gluteal region (n = 25), foot (n = 11), popliteal region (n = 6), and knee (n = 6). The most frequent histological subtypes were pleomorphic sarcoma (n = 66), myxoid liposarcoma (n = 39), myxofibrosarcoma (n = 38), and synovial sarcoma (n = 31). High-grade tumors predominated, with 140 (52.6%) classified as G3 and 90 (33.8%) as G2; grading was unavailable in 36 cases. The median tumor size was 9.1 cm (IQR: 4.7-14.0).

Oncologic treatment

RT was administered to 182 patients (68.4%), including 154 adjuvant, 29 neoadjuvant, and 20 intraoperative treatments; 20 patients received more than one RT modality. For time-to-event analyses, only patients who received neoadjuvant or adjuvant RT were included. Prior RT at the same site was documented in 16 patients (median dose, 60 Gy) and 2 patients developed radiation-associated secondary malignancies.

Chemotherapy was part of the initial treatment in 39 patients (14.7%), administered neoadjuvantly in 21 and adjuvantly in 18; an additional 25 received chemotherapy during follow-up for metastatic disease. Fourteen patients had previously received chemotherapy for unrelated conditions.

Surgical procedures and resection status

Periosteal procedures were performed in 114 patients (42.8%), and partial bone resection in 90 (33.8%). Thirteen patients required amputation after the initial operation. Resection margins were R0 in 195 patients, R1 in 51, Rx in 11, and R2 in 4; margin status was unavailable in 5 cases.

PPFs

A total of 23 PPFs occurred, excluding two patellar fractures. Fractures developed between 2 months and 97 months postoperatively. The cumulative fracture incidence was 1.8% at 1 year, 13.0% at 5 years, and 20.6% at 10 years (Figure 1A).

Figure 1 Fracture rate in months after operation based on Kaplan-Meier analysis. A: Patellar fractures; B: Arterial hypertension was associated with an increased fracture risk. HR: Hazard ratio.

Most fractures involved the femur (n = 17), followed by the lower leg (n = 5) and hip (n = 1); no fractures occurred in the foot. Among femoral fractures, nine were diaphyseal, two involved the femoral neck, and one was condylar, seven of the diaphyseal fractures occurred in the proximal third. No PPFs occurred in STS.

Risk factors for PPFs

Patients who developed PPF were older than those without fracture (median: 64 years vs 59 years; P < 0.05). Although PPF occurred more frequently in women than in men (10.3% vs 7.3%), this difference was not statistically significant (HR: 1.34; 95%CI: 0.59-3.03; P = 0.49). Two of 5 patients with osteoporosis developed PPF (HR: 4.25; 95%CI: 0.98-18.38; P < 0.05), although these estimates should be interpreted cautiously given the small sample size. Arterial hypertension was associated with an increased fracture risk (HR: 3.0 at 10 years; 95%CI: 1.27-7.07; P < 0.01; Figure 1B).

Increased fracture risk was observed in pleomorphic sarcoma, liposarcoma, and leiomyosarcoma (histological subtypes with > 10 cases; P < 0.05; Figure 2). Tumors in patients with PPF were larger than those without fracture (median: 12.0 cm vs 8.9 cm). The predictive performance of tumor size for fracture risk was assessed using ROC curve [area under the curve (AUC) of 0.61]. The optimal cut-off identified by the Youden index was 11 cm. Using a clinically relevant threshold, tumor size ≥ 10 cm was associated with increased fracture risk (HR: 2.04; 95%CI: 0.89-4.72; P = 0.08; Figure 3A).

Figure 2 Total fractures according to soft tissue sarcomas type for every sarcoma n > 10.

Figure 3 Fracture conditions in different classification scenarios. A: Fractures according to soft tissue sarcomas size; B: Postoperative fracture rate of patients with soft tissue sarcomas of the lower extremity and hip depending on procedures involving the bone; C: Postoperative fracture rate according to radiotherapy. HR: Hazard ratio.

PPF occurred in 7 of 90 patients with G2 tumors and 14 of 140 with G3 tumors (G3 vs G2: HR: 1.57; 95%CI: 0.63-3.90; P = 0.33). It occurred in 6 of 51 patients with R1 resection compared with 16 of 195 with R0 resection (44% relative increase). Neither association reached statistical significance.

Surgical technique and fracture risk

Patients undergoing periosteal stripping or partial bone resection had higher fracture rates than those without bone involvement, with cumulative fracture rates of 3.3%, 17.9%, and 27.5% at 1, 5, and 10 years, respectively, compared with 0.8%, 7.9%, and 15.6% (HR: 2.55; 95%CI: 1.11-5.83; P < 0.05; Figure 3B). When stratified by technique, 1 of 24 patients treated with periosteal stripping alone developed PPF, compared with 14 of 90 patients undergoing partial bone resection (HR: 3.7; 95%CI: 1.05-12.77; P = 0.18).

RT and fracture risk

PPF occurred in 22 of 182 patients receiving RT compared with 1 of 84 patients without RT. The 10-year fracture risk was 25.7% with RT and 1.8% without (HR: 6.5; 95%CI: 2.44-17.14; P < 0.05; Figure 3C). PPF occurred in 1 of 29 patients receiving neoadjuvant RT (3.5%) and in 20 of 154 receiving adjuvant RT (13.0%), without a statistically significant difference (HR: 2.33; 95%CI: 0.31-17.5; P = 0.21). Only neoadjuvant and adjuvant RT were included in this analysis. The median RT dose was identical in patients with and without PPF (60 Gy in both groups; P = 0.26). However, the overall dose distribution was higher in patients with PPF, as reflected by the IQRs (60-66 Gy vs 50-66 Gy). The median number of fractions was 25 (IQR: 25-30) in patients with PPF and 25 (IQR: 25-28) in those without. The median dose per fraction was 2 Gy in both groups. Most patients received IMRT (n = 71), followed by 3D-conformal RT (CRT) (n = 59) and raster scanning (n = 8). PPF occurred in 10.9% and 16.9% of patients treated with IMRT and 3D-CRT, respectively, with no fractures observed in the raster scanning group. A comparison of baseline characteristics between IMRT and 3D-CRT groups is provided in Table 1.

Table 1 Comparison of baseline characteristics between 3D-conformal radiotherapy and intensity-modulated radiotherapy groups, n (%)/median (interquartile range).

Characteristic	3D-CRT (n = 59)	IMRT (n = 71)
Age (years)	56 (45-72)	59 (47-71)
Male	39 (66)	35 (49)
Female	20 (34)	36 (51)
BMI	27.5 (24.7-32.2)	26.4 (24.0-28.3)
Max tumor diameter (cm)	11.0 (5.5-15.0)	9.8 (6.8-15.5)
Tumor size > 10 cm	30 (51)	32 (45)
R1-resection	12 (20)	15 (21)
G3-STS	37 (63)	35 (49)
Periosteal stripping	7 (12)	6 (8)
Bone resection	24 (41)	20 (28)
RT dose (Gy)	60 (50-60)	60 (50-66)
Arterial hypertension	23 (39)	27 (38)
Osteoporosis	0 (0)	4 (6)

BMI: Body mass index; STS: Soft tissue sarcomas; RT: Radiotherapy; IMRT: Intensity-modulated radiotherapy; CRT: Conformal radiotherapy.

Multivariate analysis

Multivariable analysis was performed using Cox regression with Firth’s penalized likelihood to account for the low number of events and reduce the risk of overfitting. Variables were selected based on clinical relevance and univariable time-to-event analyses. In this model, partial bone resection (HR: 3.20; 95%CI: 1.32-7.77; P = 0.007) and osteoporosis (HR: 6.01; 95%CI: 1.39-25.94; P = 0.037) remained independently associated with PPF. RT showed an increased hazard (HR: 3.16; 95%CI: 0.57-17.45; P = 0.116) but did not reach statistical significance. Age and arterial hypertension were not independently associated with fracture risk (Tables 2 and 3).

Table 2 Multivariate Firth’s penalized Cox regression analysis for risk factors.

Characteristic	Hazard ratio	95%CI	P value
Age	1.02	0.99-1.05	0.185
Arterial hypertension	1.99	0.72-5.44	0.166
Radiotherapy	3.16	0.57-17.45	0.116
Bone resection	3.20	1.32-7.77	0.007
Osteoporosis	6.01	1.39-25.94	0.037

Table 3 Multivariate Firth’s penalized Cox regression analysis for risk factors.

Characteristic	Hazard ratio	95%CI	P value
Age > 60 years	1.47	1.68-63.00	0.469
Arterial hypertension	2.36	2.32-746.40	0.103
Radiotherapy	3.20	1.78-51719005.04	0.182
Bone resection	3.27	3.83-2857.43	0.009
Osteoporosis	6.28	4.22-803808597164.85	0.014

DISCUSSION

In this large single-center cohort of patients with lower extremity STS treated with limb-sparing surgery, PPFs were a clinically relevant late complication, with a cumulative incidence of 20.6% at 10 years. Fractures occurred predominantly in the femur and often several years after treatment, consistent with prior reports describing PPFs as a delayed sequela of combined surgical and radiotherapeutic management rather than an early postoperative event[12,18,19].

The fracture incidence observed in the present study falls within the upper range reported in the literature, where cumulative rates vary widely from approximately 2% to 25%[8,12,16,18-21]. This variability has been attributed to differences in fracture definitions, extent of surgical bone involvement, RT techniques, and duration of follow-up[18,20,21]. The extended follow-up in the present cohort likely contributed to the relatively high cumulative incidence, as late fractures may be missed in studies with shorter surveillance periods[12,18,19].

RT was associated with an increased fracture risk in univariable analysis. Although it did not reach statistical significance in the multivariable model, the HR remained elevated, suggesting a clinically relevant effect consistent with prior literature. This finding aligns with multiple studies demonstrating increased fracture risk following RT for extremity STS[12,16,18-20].

Earlier studies from the pre-IMRT era reported particularly high fracture rates when RT was combined with periosteal stripping or circumferential femoral irradiation[12,20]. More recent data from IMRT-treated cohorts suggest lower fracture rates compared with historical estimates, although fractures remain a relevant complication even with modern techniques[22]. In the present study, fewer fractures were observed in patients treated with IMRT compared with 3D-conformal RT; however, subgroup sizes were insufficient to draw definitive conclusions regarding technique-specific risk reduction, consistent with prior reports[22].

Surgical involvement of bone was independently associated with fracture risk in the present cohort, with partial bone resection conferring a significantly increased hazard of PPF. This finding aligns with prior studies identifying periosteal stripping and cortical bone resection as key contributors to postoperative skeletal fragility[12,15,16,18,20]. Although periosteal stripping alone was associated with fewer fractures than partial bone resection in the present analysis, the low number of events limits meaningful comparison between surgical techniques, a limitation also noted in prior cohorts[18,20].

Tumor size was associated with fracture risk in univariable analysis but did not retain statistical significance in time-to-event modeling. ROC analysis demonstrated modest predictive performance (AUC of 0.61), and although a clinically relevant threshold of 10 cm was associated with an increased hazard, this did not reach statistical significance. These findings suggest that tumor size may reflect treatment intensity rather than serving as an independent predictor. This interpretation is consistent with prior studies and predictive models linking larger tumors to increased fracture incidence[15,16,18]. Larger tumors often require more extensive resections and higher radiation exposure, making it difficult to disentangle tumor-related effects from treatment-related factors[15]. Accordingly, tumor size likely reflects cumulative local treatment burden rather than an independent biological driver of fracture risk[15].

Patient-related factors demonstrated weaker and less consistent associations. Older age and arterial hypertension were associated with fracture risk in univariable analyses but did not retain statistical significance after adjustment, consistent with the heterogeneous and often inconclusive findings reported in prior cohorts[16,18,19,21]. Osteoporosis remained independently associated with fracture risk in multivariable analysis; however, estimates should be interpreted cautiously given the small number of affected patients and limited statistical power, a limitation also noted in earlier studies[16,18,21].

An increased fracture incidence was observed in selected histological subtypes in univariable analysis; however, histology-specific conclusions are limited by small subgroup sizes and confounding by tumor size and treatment intensity. Prior studies have similarly reported inconsistent associations between histology and fracture risk, and histological subtype has not consistently emerged as an independent predictor in multivariable analyses[15,16,18]. Consequently, histology was not retained in the multivariable modeling in the present study.

Study limitations and future directions

Several limitations warrant consideration. The retrospective design introduces potential selection and information bias, and despite the relatively large cohort, the absolute number of fracture events limits statistical power for subgroup and interaction analyses. Treatment strategies evolved over the study period, particularly with respect to RT techniques, which may introduce residual confounding. These limitations are consistent with much of the existing literature, which is dominated by retrospective single-center cohorts and heterogeneous methodologies[8,12,16,18-21].

Despite these limitations, the present study provides long-term follow-up data from a contemporary cohort and confirms, in multivariable analysis, the central role of bone-involving surgery and osteoporosis in the development of PPFs. RT showed an elevated but not statistically significant hazard. These findings are consistent with prior evidence and support careful preoperative risk assessment, judicious surgical planning when bone involvement is anticipated, and sustained long-term follow-up for patients at increased fracture risk[12,15,16,18-20,22]. Future prospective studies with standardized fracture definitions and modern RT protocols are needed to further refine risk stratification and evaluate preventive strategies, as emphasized in prior reviews[8,21].

CONCLUSION

PPFs represent a clinically significant late complication following limb-sparing surgery for lower extremity STS. In this large single-center cohort with long-term follow-up, partial bone resection and osteoporosis were independently associated with increased fracture risk, whereas RT showed an elevated but not statistically significant hazard. Tumor size was associated with fracture risk in univariable analyses but did not retain significance in time-to-event modeling. Fractures occurred predominantly in the femur and often several years after treatment, underscoring the need for sustained postoperative surveillance. These findings support careful risk stratification and multidisciplinary treatment planning in patients undergoing limb-sparing management for STS.