Preoperative coagulation biomarkers associate with survival and pulmonary embolism after surgical treatment of non-spinal skeletal metastases

Cancer, trauma, and orthopedic procedures are well known risk factors for thromboembolic complications [1‐3]. In orthopedic surgery, the rate of venous thromboembolism (VTE) after major elective lower extremity surgery with postoperative thromboprophylaxis is estimated to be 1% [4‐6]. Oncologic surgery further increases this risk, as the rate of VTE is 5% with postoperative thromboprophylaxis in abdominal cancer surgery [7]. In patients with surgically treated skeletal metastases, the incidence rate of thromboembolic complications is even higher, despite thromboprophylaxis. Their rate for deep venous thrombosis (DVT) is 2–11%; and that of pulmonary embolism (PE) 2–9%, and for lethal PE under 1% to over 3% [1, 8‐11]. The overall postoperative mortality rate is high, ranging from more than 7% during the first 30 days to between 38 and 47% at 1 year [8‐12].

Cancer leads to an imbalance in coagulation and hemostasis, which further contributes to cancer progression. The clinical manifestations of this imbalance range from DVT and PE to hemorrhagic complications [13]. Preoperative laboratory tests before major, hemostatically risky surgery usually include complete blood count, prothrombin time (PT) as a marker of vitamin K-dependent clotting factors, and activated partial thromboplastin time (APTT), which reflects the intrinsic pathway. Fibrinogen is the most abundant coagulation protein and the source of fibrin formation, and may prolong thrombin time (TT) in cancer-associated dysfibrinogenemia. Coagulation factor VIII (FVIII) and fibrinogen as acute phase reactants are associated with a risk for thrombosis and its reoccurrence, especially in cancer patients [2, 14, 15].. D-dimer depicts fibrin turnover, depending on both coagulation and fibrinolysis, and is often elevated in cancer patients, especially in association with chemotherapy-linked VTE [16].

FVIII, fibrinogen, and D-dimer have recently all been reported to improve diagnostics and aid surgical decision-making in pancreatic [17‐19] and gynaecological cancers [20]. Therefore we aimed at identifying associations between coagulation biomarkers and postoperative thromboembolic complications and survival in patients who had undergone surgery for non-spinal actual or impending pathologic fractures. We targeted the potential role of biomarkers to aid in surgical decision-making, and to consider novel thromboprophylactic strategies.

Patients were identified from two referral bone tumor centers in Finland. All were operatively treated for non-spinal skeletal metastases due to actual or impending pathologic fractures between March 20th, 2016 and November 14th, 2020. An actual fracture has occurred, and an impending fracture has a degree of bone disruption which warrants prophylactic surgical stabilization to prevent a fracture, intractable pain, or the loss of ambulatory ability. Data from all consecutive patients were collected retrospectively from medical records. The study was approved by the institutional ethical review boards (HUS/234/2020, 14.1.2020 Helsinki, Finland).

All blood and plasma samples were collected preoperatively as part of routine diagnostic laboratory examinations, and analyzed at Helsinki University Hospital Laboratory HUSLAB, and Tampere University Hospital Laboratory FimLab. The following variables were captured: 3. A traditional subcutaneous low-molecular-weight heparin (LMWH) dose, either enoxaparin 40 mg or tinzaparin 4500 IU once daily was administered for 30 days. The laboratory variables included blood cell counts (of red blood cells, leukocytes and platelets), creatinine, and a coagulation panel with PT, TT, APTT, fibrinogen, FVIII, antithrombin, and D-dimer. The analysis covered both the whole group, and the most common primary diseases (multiple myeloma, breast, kidney and lung cancers).

In this retrospective cohort study, 100 patients with 113 pathologic fractures treated operatively were studied to identify potential coagulation biomarkers for postoperative survival and complications. Preoperative fibrinogen and FVIII have been proposed as biomarkers and contributors for progressive primary malignant disease [18, 21, 22], and our findings further extend these observations to survival in patients having pathologic fractures. Since coagulation activity and fibrin generation may remain non-overt, i.e. without overt thrombosis, the preoperative coagulation biomarkers could reveal the underlying enhanced interactive mechanisms between coagulation and cancer. Therein, our results highlight the need for an overall improvement in anticoagulant strategies in patients who undergo surgery of pathologic fractures, including the upper extremity. Beyond this strong cancer-associated coagulation activity, PE associated only with fibrinogen levels only in its earlier appearance post surgery.

The role of the biomarkers of coagulation activity in patients with pathologic fractures has not previously been studied. In the study by Mattila et al. on biomarkers of coagulation in pancreatic tumors, the preoperative acute phase reactants fibrinogen and FVIII distinguished cancer from benign tumors [17, 18, 20]. Our data on pathological fractures showed that these coagulation biomarker levels are significantly increased in all patients. Indeed, fibrinogen levels exceeding normal ranges (4 g/l) were associated with decreased postoperative survival, and fibrinogen levels exceeding 5 g/l also increased the risk for postoperative PE. The higher the fibrinogen, the stronger the inflammatory reaction, and the faster the fibrin formation, the poorer its resolution [23].

Neither FVIII nor D-dimer were associated with postoperative PE, but elevated FVIII did indicate decreased survival. In addition, high FVIII may be a universal predictor of postoperative survival for patients with pathologic fractures, as it displayed limited variation between patients with different primary diseases. High FVIII was associated with overall mortality also in a larger population study investigating inflammatory diseases [24]. FVIII derives from endothelial activity, is carried by released von Willebrand factor, and enhances thrombin generation. D-dimer is the joint marker of both coagulation activity and fibrinolysis, and cancer has been associated with both enhanced coagulation and aberrant fibrinolytic system. Local tissue remodeling occurs at sites where fibrinolytic degradation takes part in tumor development and progression. Thus, the systemic fibronolytic balance is likely to be altered and fibrin resolution impaired in advancing cancer [25].

Literature on PE among patients with surgery for pathologic fractures is scarce, and the rates differ markedly between 2 and 8.5% [1, 8, 9, 12, 26]. When compared with the reported rates reported, our PE rate is slightly higher. However, we did not observe the adverse effects of PE on overall survival, which have been observed in other larger studies [8, 9]. Further, because on average PE occurred on post-operative day 36, the 1 month duration of thromboprophylaxis seems insufficient as these patients are in a procoagulant state, likely causing rebound activation of coagulation upon cessation of LMWH, while the preoperative role of fibrinogen and FVIII only gets stronger (OR 8.9) later on, for instance at 2 years.

Our observations may add to other scoring systems predicting survival when considering the method of surgery, since aggressive massive resection and large reconstruction could be avoided in those patients having limited benefit from the procedures. A combination of FVIII and fibrinogen may serve as a biomarker for postoperative survival, as together they could identify the patients who would be poor survivors. Although D-dimer is a well-known predictor of PE, D-dimer was elevated in our study in all patients with pathologic fractures and did not associate with survival or PE. The combination of fracture, exposure of blood to bone and cancerous tissue and surgical trauma may consume plasmin activity, and therefore fibrin may be inadequately degraded, as several metabolic states reflect on fibrinolysis [27].

Research on thromboembolism after orthopedic procedures of the upper extremities is limited. The rate of postoperative thromboembolism after humeral fractures, even without thromboprophylaxis, has been considered low [28], and the routine use of thromboprophylaxis following upper extremity trauma has not been recommended. A previous meta-analysis reported a rate of postoperative DVT of less than 1% and no postoperative PE in patients who underwent surgery for primary skeletal tumors of the upper extremity [1]. One study showed a similar rate of thromboembolic complications after intramedullary nailing of upper and lower extremity pathologic fractures of long bones [26]. Along systemic blood-bone interactions, such as coagulation activity, we show that the incidence of postoperative PE is similar irrespective of the site of skeletal metastases and fractures and suggest that thromboprophylaxis should be administered to all pathological fractures to avoid cancer-associated thrombosis.

Several limitations may have influenced the results of this cohort study. As the study design is retrospective and observational, selection bias may be an issue. The patients were, however, collected from two centers with centralized patient information systems, and the documentation of the patient information is standardized. Moreover, the treatment protocols, laboratory services, and indications for surgery of these two university clinics are similar. The patients had different primary diseases at different stages. Also, a minority of patients used anticoagulants preoperatively for other indications. The follow-up times for each patient varied. Fibrinogen, as an indicator of increased risk for PE, showed significant results in Kaplan-Meier analysis but not in Cox regression analysis, and these results need to be verified in larger patient groups. The potential biomarker use of fibrinogen to identify the need for a more aggressive and longer or a different thromboprophylactic protocol also demands further studies.

The results of our study establish fibrinogen and FVIII, acute phase reactants promoting coagulation, as suitable preoperative biomarkers for postoperative survival and complications after the operative treatment of pathologic fractures. Overall, fibrinogen appears to be a more sensitive marker for postoperative survival and PE than FVIII, which was very high on average (exceeding 250 IU/dL). Fibrinogen, however, varies between patient groups with different primary diseases, and therefore the combination of fibrinogen and FVIII captured the outcome better across all cancer types. In addition, our results highlight the need for novel improved thromboprophylaxis strategies due to the high incidence of PE. Specifically, patients with metastatic fractures of the upper limb should be better protected from VTE complications. A randomized study setup for both pre- and postoperative compared with only postoperative anticoagulation or thromboprophylaxis would be of interest.