Risk Factors for Early Postoperative Morbidity and Mortality following Extremity Metastatic Pathologic or Impending Fracture Fixation

Background As cancer survivorship continues to improve, the perioperative morbidity and mortality following surgical treatment of metastatic bone disease become an increasingly important consideration. The objective of this study is to identify risk factors for early postoperative complications and mortality following extremity prophylactic fixation and pathologic fracture stabilization. Methods A single-centre retrospective review of 185 patients (226 total surgeries) who underwent prophylactic fixation or pathologic fracture stabilization for extremity metastatic bone disease between 2005 and 2020 was performed. Skull, spine, pelvic, and revision surgeries along with diagnosis of primary bone neoplasm were excluded. Univariate, multivariate, and subgroup analyses were performed to identify predictors and independent risk factors for 30-day postoperative morbidity and mortality. Results Primary cancers included lung (n = 41), breast (n = 36), multiple myeloma (n = 35), prostate (n = 16), lymphoma (n = 11), renal cell carcinoma (n = 10), and other (n = 36). The 30-day postoperative complication and mortality rates were 32.30% (n = 73) and 17.26% (n = 39), respectively. The most common complications were pulmonary-related, cardiac events, surgical site infection, sepsis, and thromboembolism. Pathologic fracture, presence of extra-skeletal metastases, longer surgical duration, and blood transfusion requirements were associated with 30-day postoperative complications overall. A past medical history for cardiac disease was also associated with systemic but not local surgical complications. Pathologic fracture, presence of extra-skeletal metastases, lung cancer, blood transfusion requirements, and increasing pack-year smoking history were predictors for 30-day mortality. In the multivariate analysis, pathologic fracture (p=0.016) and presence of extra-skeletal metastases (p=0.029) were independent predictors of complications. For mortality, pathologic fracture (p=0.014), presence of extra-skeletal metastases (p=0.0085), and increased blood transfusion requirements (p=0.048) were independent risk factors. Conclusions The findings of this study provide additional guidance for perioperative risk assessment and patient counselling. Additionally, improving clinical assessment tools to identify and quantify patients at risk of pathologic fractures becomes increasingly important given the significant morbidity and mortality associated with pathologic fracture treatment.


Introduction
Te incidence of metastatic bone disease is expected to continue to rise as the life expectancy of cancer patients continues to increase resulting from advances in screening techniques as well as local and systemic treatments.
Te reported one-year survival rate following a pathologic fracture is between 30% and 70%, while the 2-year survival rate is less than 30%, all cancer aetiologies included [2,6,[10][11][12][13][14]. Terefore, the aim of surgical intervention is immediate pain relief, restoration of function, and improved quality of life often in the setting of end-of-life care, while minimizing perioperative complications [1,3,5,[14][15][16].At present, increased emphasis has been placed on creating models to predict symptomatic skeletal events from the time of cancer diagnosis and to identify prognostic factors for overall survival.However, there are limited data exploring the risk factors for early postoperative morbidity and mortality from the time of impending or pathologic fracture diagnosis, which is usually when patients are seen by an orthopaedic surgeon for the frst time.Tis has great implications in clinical decision-making, perioperative risk assessment, as well as patient and family counselling [1,6,8,12,17,18].
Te objective of this study is to identify risk factors for short-term complications and mortality in the surgical treatment of extremity metastatic bone disease.Ultimately, the aim is to provide decision-making guidance and to enhance perioperative risk assessment.

Methods
A retrospective review of consecutive patients who underwent surgical treatment for impending and/or pathologic fractures related to extremity metastatic bone disease from January 1, 2005, to December 31, 2020, was performed.Primary bone neoplasm, revision surgeries, nonsurgical management, and treatment of skull, spine, pelvis, or nonskeletal metastatic disease were excluded.All patient consultation and surgical interventions were performed at a single tertiary centre with regional cancer care designations.
Te data were independently collected by two authors (PQW, BPC) from the institution's electronic medical record.Discrepancies were reviewed until consensus was reached.Primary endpoints were 30-day postoperative complications and mortality.A complication was defned as a new or worsening condition requiring additional medical or surgical treatment.An impending fracture was defned as a metastatic bone lesion without a visible fracture line, with intact height and bone alignment, and which underwent prophylactic fxation [19].For risk factor assessment, variables of interest included age, sex, primary cancer, time (in months) between cancer diagnosis and surgery, chemotherapy and radiation therapy history, fracture location, prophylactic versus fracture stabilization, type of fxation, type of anesthesia, surgical duration, metastatic burden, medical comorbidities, smoking history, preoperative (or prefracture) ambulatory status, and transfusion requirements.Metastatic burden was dichotomized into either bone metastasis alone or presence of extra-skeletal metastases.Surgical location was categorized as upper extremity, lower extremity, and multiple limbs given the propensity to metastasize to both the humerus and femur, and less frequently to other bones [12,19].Medical comorbidities needed to be preexisting conditions and unrelated to the cancer diagnosis and/or treatment.Te smoking history and blood transfusion requirements were assessed using categorical values (yes/no) as well as continuous variables (pack-year smoking history and number of blood product units transfused, respectively).Where patients underwent multiple separate surgeries for diferent limbs within 30 days and sustained a complication or incurred death, these data were linked only to the most recent surgery and not the previous interventions as the patient would have been expected to be optimized for surgery.
Patient demographics, clinical characteristics, 30-day morbidity, and mortality were summarized using descriptive statistics.Categorical variables were presented as frequencies and percentages, whereas continuous independent variables were reported as mean, standard deviation, and range.Postoperative morbidity and mortality were assessed separately using univariate and multivariate analyses.Associations between each variable and outcome were identifed using chi-squared or Fischer's exact test for categorical values.A binary logistic regression model was used for continuous independent variables.A multivariate regression analysis was subsequently performed to identify independent risk factors for postoperative complications and mortality.Variables that were statistically signifcant in the univariate analysis were included in the multivariate model and were reported as odds ratio (OR) with 95% confdence interval (CI).An additional subgroup analysis was performed to identify specifc risk factors for local surgical and systemic complications separately.Complications occurring outside the surgical feld were considered systemic.Statistical signifcance was set at p value less than 0.05.Statistical analysis was performed using GraphPad Prism 10 (San Diego, California, USA).
Approval from the institution's research ethics committee was obtained prior to initiating data collection.Tis study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [20].
Te subgroup analysis assessing risk factors stratifed by local (n � 17) and systemic (n � 56) complications showed that pathologic fracture (p � 0.012), presence of extraskeletal metastases (p � 0.0051), cardiac medical history (p � 0.049), and increasing amounts of blood transfusion units (p � 0.032) were associated with systemic complications.However, no statistically signifcant risk factors were identifed for local surgical complications alone.

Discussion
Te objective of this study was to identify the risk factors for short-term postoperative morbidity and mortality to guide clinical decision-making and enhance perioperative risk assessment in the treatment of impending and pathologic fractures of the extremity due to metastatic bone disease.Several risk factors were identifed, of which pathologic fractures, presence of extra-skeletal metastases, and increased blood product requirements were independently associated with morbidity and/or mortality.A cardiac past medical history was also a signifcant risk factor for systemic complications in the subgroup analysis.Te most common complications included pulmonaryrelated morbidity, cardiac events, surgical site infections, sepsis, and thromboembolism.Te treatment of pathologic fractures, similarly reported in previous studies, has signifcantly worse outcomes compared to prophylactic stabilization and is an independent negative prognostic factor in the treatment of extremity metastatic disease [11,12,21,22].Previous studies demonstrated that long bone pathologic fracture fxation compared to prophylactic stabilization resulted in a signifcantly higher rate of 30-day major medical complications, longer hospital stays, and higher likelihood of patient discharge into a secondary health care facility rather than home discharge [19,23] and is associated with an increased 1-year mortality rate [19,24].Consequently, identifying patients at a higher risk of pathologic fracture and advocating for treatment of impending rather than completed fractures, all factors considered, is becoming increasingly important.At present, assessment tools quantifying true risks of pathologic fractures are still limited.Mirel's scoring system is still the most widely used screening tool for predicting the risk of pathologic fracture.However, its low specifcity rate (35%) leads to a high false-positive rate, thereby leading to overtreatment of extremity metastatic bone disease if decisions were made solely based on Mirel's score.Additionally, previous studies have demonstrated a moderate interobserver agreement only.Computed tomography-based bone assessment models have shown promising results in predicting the risks of pathologic fractures.However, given its complexity, it is still limited in availability and lacks standardization [22,23,[25][26][27][28].
Prognostic predictive models such as PATHFx, 2013-SPRING, Optimodel, and SORG (the Skeletal Oncology Research Group) have been developed over the years to assess overall survival prognosis following treatment of metastatic bone disease.However, they do not quantify the risk of completed pathologic fracture.While helpful, these clinical algorithms have not been validated as stand-alone decision-making tools for management of patients with metastatic bone disease, rather, a multidisciplinary approach remains the gold standard.Furthermore, these tools are not yet used routinely, even amongst orthopaedic oncologists.As a result, emphasis should be placed on developing precise and accurate clinical prognostic tools both in quantifying risks of pathologic fracture as well as prognosis to make better informed decisions.Tese tools should be simple to apply and available, not only for clinicians in tertiary oncology referral centres, but also for secondary centres who also treat these patients.Tis is ever-more important now that patients survive longer and an increasing number of symptomatic skeletal events are expected to occur [22,26,[29][30][31][32].
Similar to this study, reported acute postoperative complications and mortality are observed in up to 31% and 19%, respectively [12,19,21,29].Te most common reported postoperative complications are respiratory complications, cardiac events, surgical site infections, sepsis, and thromboembolism [18,19,23].Common prognostic factors for morbidity and/or mortality previously published include rapidly growing primary tumours, lung cancer, disseminated disease, age (older),  International Journal of Surgical Oncology blood transfusion requirements, lower extremity location, and several laboratory values [18,19,29].Unlike previous articles, in the current study, tumours were not diferentiated by slow, moderate, and rapid growth, although lung cancer, known for its aggressive potential, was a risk factor in the univariate analysis for mortality, and metastatic burden (advanced disease) was an independent factor for both morbidity and mortality.Laboratory values were categorized as a complication only if they required additional treatment, thus accounting for the discrepancies from previous data.In addition, perioperative albumin levels were not routinely performed.It has been postulated that blood transfusion alone may not be a prognostic factor, rather it is the low preoperative hemoglobin level and perioperative anemia that afect patient outcomes, hence potentially why transfusion requirement (yes/no) was not an independent risk factor in this study, but mortality was afected by the increasing amount of blood transfusion requirements [11,18].A cardiac medical history as a risk factor for systemic complications was identifed in the subgroup analysis but not in the univariate analysis, although it did trend towards, but did not reach signifcance.Tus, cardiac disease likely has a certain efect on patient morbidity, and the discrepancy could be explained by the low sample size and heterogeneity in data.Tis study did not show an increased risk of morbidity for lower extremity compared to upper extremity surgery, unlike observations made from previous studies.First, this could be explained by diferences in patient demographics, cancer status, type of surgery, and how complications were defned.Additionally, lower extremity surgeries often require longer hospital stays and more intense rehabilitation compared to the upper extremity.However, most studies assess long-term morbidity, likewise with implant-related failures and complications.Consequently, morbidity related to lower extremity surgery and implant failures likely become more signifcant beyond the initial 30-day period [19,[33][34][35][36].
Tere is conficting evidence as to whether older age is an independent risk factor in long-term outcome studies.Previous pooled analyses proposed some studies showing age as an independent risk factor, while other studies did not reproduce the same results [11].Diferences can be infuenced by the cancer types and whether they are assessed separately or together.For example, age afects survival in  8 International Journal of Surgical Oncology breast, prostate, renal cell, and thyroid cancers when analyzed independently [11,[37][38][39][40][41].However, studies with pooled primary cancer diagnoses, often do not demonstrate age as a prognostic factor [12,42].Te current study did not demonstrate associations between prior chemotherapy and prior radiation with shortterm morbidity or mortality.Similarly, Bindels [19] did not fnd correlations between 30-day complications and previous systemic therapies or radiation to the afected bone [19].However, given that disease dissemination and metastatic burden afect outcomes, it is more likely that the actual response to systemic treatment will have a greater impact on overall survival, especially if it can delay disease progression [7,43].
Tis study has several limitations stemming from its retrospective design, heterogenous patient demographics and characteristics, and its reliance on electronic medical record.With a relatively small sample size, not all variables could be independently analyzed.As a result, cancer types and complications were pooled together instead to increase power.Furthermore, in the literature, there is no uniform defnition and data collection method for complications, thereby making direct comparisons difcult [19].Certain types of surgical procedures from diferent locations were also combined to increase subgroup sample size, notably for joint reconstruction surgeries.Total joint replacements and megaprosthetic reconstructions were pooled with hemiarthroplasties of the shoulders, hips, and knees although their risk profles difer, thereby increasing the risk of bias.Other potential confounding factors such as nutritional status, albumin levels, and various laboratory bloodwork could not be independently analyzed due to lack of data.Rather, only the reported abnormal laboratory values that required additional treatment were categorized as a complication.Additionally, information regarding oncologyrelated patient performance status, an important prognostic factor for survival, was not available in the chart review.Terefore, this study assessed ambulatory status as a surrogate measure.While preoperative ambulatory status did not appear to afect the short-term results, ambulatory status alone is not an adequate representation of functional status [29].

Conclusions
Tis study highlights the various complications and risk factors associated with early postoperative morbidity and mortality following surgery for extremity metastatic bone disease.After adjusting for potential confounders, pathologic fractures and the presence of extra-skeletal metastases were independent risk factors for morbidity and mortality.Meanwhile, increased blood transfusion requirement was a negative prognostic factor for mortality.Finally, as patient survivorship continues to improve, there is an increasing role for prophylactic treatment given their superior clinical outcomes and the morbidity associated with pathologic fracture surgery.Accurately identifying patients at a risk for pathologic fracture is paramount.

Table 1 :
Demographic data and clinical characteristics.

Table 2 :
Details of surgical procedures.
SD: standard deviation, * surgery time is measured from initial incision until fnal wound dressing is placed.**Multiple limb procedures in one surgery.aStatisticallysignifcant risk factor for systemic complication.Te bold values indicate statistically signifcant.6InternationalJournal of Surgical Oncology
SD: standard deviation, * surgery time is measured from initial incision until fnal wound dressing is placed.* * Multiple limb procedures in one surgery.Te bold values indicate statistically signifcant.

Table 5 :
Multivariate analysis for 30-day complications and mortality.