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

Characteristics of patients with operatively treated pathologic fractures

A total of 113 pathologic actual (n = 94, 83%) or impending (n = 19, 17%) fractures were treated surgically in 100 patients (Table 1).

Table 1 Characteristics of patients with upper and lower limb metastases

Mean age of the patients at the time of operation was 69 years (range, 32–96 years), 54 patients were female and 46 male, and the most common primary diseases were multiple myeloma, breast, kidney, and lung cancers (Table 1). Thirty-seven patients had only skeletal metastasis, and 36 patients had both skeletal and pulmonary metastases, 8 patients had skeletal and liver metastases, and 16 patients had skeletal, pulmonary, and liver metastases. Mean CCI was 8.4 (median 8.0, range 6–13). Eleven patients received radiotherapy and 28 chemotherapy.

Mean postoperative survival was 17 months (median 6 months), and the survival at 1 month was 87%, at 3 months 61%, at 6 months 45%, at 1 year 39%, and at 3 years 14% (Table 2). Any major bleeding complications, such as requiring revision surgery or other intervention did not occur, neither minor bleeding complications were documented.

Table 2 Fibrinogen, FVIII, and D-dimer by survival timeBlood cell counts

We examined the differences in the routine laboratory variables, including complete blood cell counts and coagulation screens, between the most common primary malignant diseases (multiple myeloma, breast, kidney and lung cancers) compared with the other cancer types (Table 3). In general, basic blood cell count did not vary between patients or primary disease groups, except leukocyte count in patients with kidney and lung cancer, red blood cell count in patients with breast cancer, and platelet count in patients with multiple myeloma. One patient with multiple myeloma and subtrochanteric femoral fracture had a low hemoglobin count at admittance (64 g/l), partly caused by the fracture. The hemoglobin count increased to 98 g/l by red cell transfusion prior to the operation. This same patient also presented with a prolonged thrombin time of over 140 s, signifying cancer-associated dysfibrinogemia.

Table 3 Laboratory variables in all patients and patients with PE and some cancer typesFibrinogen

Mean fibrinogen was high in all patients (reference range 2–4 g/l, mean in all 5.5 g/l). Compared with the other cancer types, fibrinogen levels were the lowest in patients with multiple myeloma (p = 0.043) and breast cancer (p = 0.013), whereas the highest in patients with kidney cancer (p = 0.025) (Table 3).

APTT, FVIII, and D-dimer

APTT (reference range 28–37 s, mean 33 s) was shorter in multiple myeloma (p = 0.046) and breast cancer (p = 0.001), but longer in kidney cancer (p = 0.033) than in other patients.

Overall, FVIII and D-dimer levels were quite similarly elevated (Table 3). In all patients, mean FVIII was 250 IU/dl (reference range 60–160 IU/dl), and mean D-dimer was 2.7 mg/l (reference range < 0.5 mg/l), without differences between primary disease groups.

Incidence and timing of postoperative PE in patients with upper and lower extremity fractures

The overall incidence of PE was 12% (12/100 patients) (Table 4). PE occurred equally in patients with lower extremity fractures (11% (n = 10/92)) and upper extremity fractures (13% (n = 2/15)). Eleven patients had an actual pathologic fracture and one had an impending pathologic fracture. Of all patients with PE, six (50%) had only skeletal metastases, four (33%) had skeletal and pulmonary metastases, one (8%) had skeletal and liver metastases, and one (8%) skeletal, pulmonary, and liver metastases. Two patients with only skeletal metastasis had actual humeral fractures, one patient with multiple myeloma and the other with lung cancer.

Table 4 Characteristics of patients with postoperative PE

On average, postoperative PE in 12/100 patients occurred at the postoperative day 36 (range 0–141 days). The occurrence of PE was not associated with primary disease, number of tissues with metastases, or fracture type. An intraoperative saturation drop was recorded in four patients with PE. In one patient PE manifested on the first postoperative day. The patient had a history of DVT and preoperative anticoagulation, and two metastases were operated on the same day: a pathologic femoral neck fracture and a contralateral impending femoral diaphyseal fracture, while all others had one fracture operated at a single setting.

Records of postoperative thromboprophylaxis were available for 95 patients. While 92 patients received thromboprophylaxis, three patients with upper limb fractures were not administered with thromboprophylaxis. Of the three patients who did not get thromboprophylaxis, one manifested with postoperative PE.

Two patients died of overt PE. One patient with lung cancer and humeral pathologic fracture died at 63 days. The other patient with breast cancer and a distal femoral fracture died at postoperative day 9, showing already an intraoperative saturation drop. The first patient with upper limb fracture did not have either pre- or postoperative thromboprophylaxis, but the other patient had both forms of prophylaxis. All the other patients developed PE under postoperative thromboprophylaxis, and seven out of 12 patients also had preoperative thromboprophylaxis. In this dataset, however, thromboprophylaxis did not signify the onset of PE (Kaplan-Meier log-rank p = 0.339). Moreover, the occurrence of PE did not influence postoperative survival either in univariate or multivariate models (Kaplan-Meier, log-rank p = 0.998, Cox analysis (RR 0.825, CI 95% 0.442–1.538)).

Fibrinogen and FVIII as biomarkers of postoperative survival and PE

The results of preoperative fibrinogen and FVIII as biomarkers for postoperative survival were significantly predictive. Mean postoperative survival in patients with fibrinogen at normal range (n = 18, 2–4 g/l) was 34 months (median 35 months), but it was less than half (15 months) in patients whose fibrinogen levels were higher than the upper normal limit of 4 g/l (n = 69, median 6 months) (log-rank p = 0.009) (Fig. 1A).

Fig. 1figure 1

Kaplan-Meier survival analysis showing postoperative survival (months) relative to coagulation variables A Kaplan-Meier survival analysis of all patients, comparison of survival between patients with fibrinogen > 4 g/l and ≤ 4 g/l (log rank p = 0.009). B Kaplan-Meier survival analysis of all patients, comparison of survival between patients with FVIII > 326 IU/dl and ≤ 326 IU/dl (log-rank p = 0.002). C Kaplan-Meier survival analysis of all patients, comparison of survival between patients with both fibrinogen > 4 g/l and FVIII > 326 IU/dl, and patients with fibrinogen ≤4 g/l and/or FVIII ≤325 IU/dl (log-rank p = < 0.001). D Kaplan-Meier survival analysis of all patients, comparison of survival between patients with D-dimer > 0.5 mg/l and ≤ 0.5 mg/l (log-rank p = 0.905)

In general, FVIII was high (mean 250 IU/dl), and the level of the highest interrange quartile 326 IU/dl was chosen as the cut-off level for analysis. Mean postoperative survival in patients with FVIII levels below 326 IU/dl (n = 47) was 22 months (median 21 months), while only 7 months (median 1 month) in those with FVIII exceeding 326 IU/dl (Q3) (n = 15, log-rank p = 0.002) (Fig. 1B).

The combination of fibrinogen exceeding 4 g/l and FVIII exceeding 326 IU/dl predicted postoperative survival. Thus, in patients with both or either variable below the threshold (n = 55), the mean survival rate was 22 months (median 22 months), whereas it was only 7 months in those whose both variables exceeded the reference ranges (n = 15, median 1 month) (log-rank p < 0.001) (Fig. 1C).

Overall, these laboratory variables depicted significant results in the Kaplan-Maier analysis, and were subjected to further analysis with the Cox regression model and multivariate logistic regression analysis. Again, the multivariate Cox regression model showed significant results for both fibrinogen and combination of fibrinogen and FVIII after controlling for potentially relevant confounding variables (CCI, chemotherapy, radiotherapy, age over 65 years, hypertension, fracture, Table 5, additional file 1). Similarly, in logistic regression analysis the combination of high fibrinogen and FVIII was independently associated with survival longitudinally over 1 month (OR 8.3, 95% CI 1.7–40.6, p = 0.009), 3 months (OR 6.2, 95% CI 1.8–21.3, p = 0.004), 6 months (OR 3.6, 95% CI 1.0–12.7, p = 0.048), 1 year (OR 6.8, 95% CI 1.4–32.9, p = 0.018), and over 2 years (OR 8.9, 95% CI 1.1–72.8, p = 0.041). However, high D-dimer did not predict survival either in multivariate or univariate analysis (Fig. 1D, Table 5).

Table 5 Cox regression survival analysis of fibrinogen and FVIII levels and their combination after controlling confounding variables

In contrast to survival, none of the preoperative biomarkers (fibrinogen, FVIII, or D-dimer did not predict) the overall occurrence of postoperative PE. Of the 12 patients who developed postoperative PE, fibrinogen was available in 10 patients, FVIII in 7 patients, and D-dimer in 11 patients. Fibrinogen predicted the earlier timing of PE at levels which exceeded 5 g/l (log rank p = 0.022) (Fig. 2A), while Cox regression analysis did not provide significance in risk (RR 0.019, 95% CI 0.000–10.7, p = 0.221). Those patients whose fibrinogen level was 5 g/l or less (n = 2) had PE later at 76 days (mean) (median 65 days) than the patients whose fibrinogen exceeded 5 g/l (n = 8) whose PE occurred at 15 days (median 1 days) (Fig. 2B). PE did not influence postoperative survival.

Fig. 2figure 2

Timing of postoperative PE in relation to fibrinogen levels. A Kaplan-Meier survival analysis of patients with PE, patients with fibrinogen > 5 g/l manifested with PE earlier than patients with fibrinogen ≤5 g/l (log-rank p = 0.022). B Timeline of 12 patients with PE. Each line represents an individual patient and their postoperative survival, and circles indicate onset of PE. White circles indicate patients with fibrinogen > 5 g/l, black circles indicate patients with fibrinogen ≤5 g/l, gray circles patients with unknown fibrinogen level.

留言 (0)

沒有登入
gif