Surgeons calculating the risk of pathologic fracture are likely to focus most of their attention on the appearance of lesion of plain radiographs. However, several other factors need to be taken into consideration.
Many different criteria have been suggested including the size of the lesion; the type of cancer that metastasized to bone; the location of the metastatic lesion; pain due to the lesion; whether the lesion is lytic or blastic; irradiation of the lesion; and the use of biomechanics to predict fracture. This learning module reviews the commonly used methods and critically assesses their accuracy.
Relationship of Lesion Size to Fracture Risk:
Beals and Snell (A) published highly influential works in 1956 and 1961. Their work dealt only with patients with breast cancer and only with lesions in the femur. Of the 19 fractures that occurred in their first series, they found that 58% of these fractures were predictable using the following criteria: Presence of a metastatic lesion 2.5 cm in size or larger involving the femoral cortex or presence of a defect of the same size in any location that caused pain to the patient. These criteria were used in a second series and patients at risk for fracture underwent prophylactic fixation of the diseased bone. This treatment reduced the incidence of fracture from 32% to 9%.
Parrish and Murray (J,E) used these criteria as indications for prophylactic fixation when performing their studies on treatments of pathologic fractures. They found that using these criteria led to decreased fractures and improved the quality of life of their patients. In 1973, Fidler(B) retrospectively studied 19 patients with pathologic fractures of the femur. He found that 100% of patients with greater than 50% cortical involvement developed a fracture. Based on this data, he recommended that patients with involvement of over half of the cortex should undergo surgery to stabilize the bone. In 1981, Fidler (D) published another retrospective study of 66 patients with 100 metastases in the long bones. His results corroborated his previously reported indications for prophylactic fixation. He found that when greater than 75% of the cortex is destroyed the incidence of fracture is 80%. When less than 50% of the cortex is involved, the incidence is only 2.3%.
Zickel and Mourandian(P) studied 34 patients with lesions in the proximal femur. They concluded that involvement of even small parts of the cortex in the subtrochanteric region places the femur at high risk for fracture and warrants prophylactic fixation. According to their results, size of the lesion did not correlate with risk of fracture. Keene et al supported this conclusion as he found that all the measureable lesions that fractured had a similar extent of cortical involvement as those that did not fracture ( C).In 1982, Harrington (4) recommended intramedullary nail fixation when there is greater than 70% destruction of the cortex (G).
In 1989, Mirels (1) developed a scoring system to quantify the risk of pathologic fracture based on a retrospective study of 78 irradiated metastatic bone lesions. Unlike all the previous studies, Mirels combined four different features of bone lesions in an attempt to create a more reliable risk assessment (See Table 3).
His system assigned points to the following four variables: the location of the lesion (upper limb, lower limb, peritrochanter); the degree of pain caused by the lesion (mild, moderate, severe); the type of lesion (lytic, blastic, mixed); and the size of the lesion (<1/3, 1/3-2/3, >1/3). Adding the points from each category determines the score. His data indicated that a score of less than or equal to 7 out of 12 is indicative of a lesion not at risk for fracture. A score of 8 out of 12 is associated with a 15% risk for fracture. The risk of fracture is 33% in patients with a score of 9. Mirels concluded that a score of 9 or greater should be used as an indication for prophylactic fixation. Mirels found that the combined score was a more accurate predictor of fracture than any of the four factors used separately. Pain and lesion size were more accurate predictors than type of lesion or site of lesion. Confirming Fidler's conclusions, Mirels found that the rate of fracture was only 5% when the lesion was less than two thirds of the diameter of the bone, but increased to 81% when the lesion was greater than two thirds of the diameter of the bone.
Accuracy of Lesion Size Measurement:
Several authors have demonstrated the limitations of using size-based criteria alone. In 1986, Keene et al (C) performed a retrospective study of 203 female patients with 516 metastatic breast cancer lesions that were located in the proximal femur. They showed that 57% of the metastases could not be accurately measured from plain radiograph because they lacked a clear border between the lesion and normal bone. Size-based criteria may not be applicable to bony lesions where the cortex cannot be effectively measured, such as the spine, ribs, and pelvis (N).
Hipp et al also stated that characteristics of both metastatic bone lesions and physician observers lead to a very high degree of error and variability in the measurement of lesions (5). Two physicians reading the same radiograph and applying the same criteria might come to very different conclusions about the need for prophylactic fixation, with potentially disastrous consequences for the patient. In fact, Hipp et al found that experienced orthopedic oncology surgeons could not consistently predict strength reductions or load bearing capacity from radiographs or CT films (5). Therefore, another method needs to be found to determine factor of risk that would be easier for orthopedic surgeons to use.
Relationship of Lesion Location to Fracture Risk:
Of the long bones in the peripheral skeleton, the femur is the most common site for metastases followed by the humerus (R,U). According to Knutson et al, 88.4% of all long bone metastasis secondary to breast cancer involved the femur (R). Within the long bones, the proximal part is most likely to be affected, especially the peritrochanteric region of the femur (1, U,d). While metastases to the long bones account for less than 20% of all fractures (O), Harrington found that over half of these pathologic long bone fractures occurred in the proximal femur (G). Some authors have suggested that the femur is more likely to sustain a pathologic fracture than other long bones (A,R,T,U,V). Dijkstra et al states that 25% of all long bone metastases fracture, but the proximal femur has an incidence of 40-60% (U).
However, Fidler (D) found no difference between the rate of fracture in upper limb lesions versus lower limb lesions. Mirels also found that the peritrochanteric region, while the most common area of the femur to develop metastases, is not any more likely to fracture than other sites (1). Taking into consideration the conflicting information in the literature, it is the authors opinion that the orthopedic surgeon treating metastatic skeletal lesions should have a relatively low threshold for prophylactic fixation of proximal femur lesions. Any lesion between the lesser trochanter and the femoral head causing functional pain or larger than 2.5 cm should be fixed prophylactically. Pathologic fractures in this location produce serious morbidity. The operative procedures applicable to the proximal femur are familiar to all practicing orthopedic surgeons. The benefit of prophylactic fixation significantly outweigh the risks of surgery.
Lytic vs. Blastic Lesions While bone formation and destruction occur simulataneously in most metastatic cancers, usually one predominates over the other. Mirels (B) and others (4,f,P,N) have found that lytic lesions did have a higher risk for fracture. Mirels found in his study that none of the blastic lesions fractured, but 32% of the mixed lesions and 48% of the lytic lesions did. He theorized that lytic lesions were a result of a more advanced process of bone resorption . On the other hand, Hipp et al (g) found that even though blastic lesions do increase bone density, they do not change bone strength and they decrease the stiffness. Lytic lesions decrease both strength and stiffness of the bone.
Biomechanical Modeling of Fracture Risk Biochemical testing and computer modeling have contributed to the understanding of fracture risk. Hipp et al (5) discusses in vitro experimentation as an alternative to using clinical and radiographical data to predict pathologic fractures. Studies have shown that even small cortical defects can significantly reduce the strength of the bone (h,j,m). Brooks et al found that drill holes as small as 2.8 mm or 3.6 mm in the femoral mid-shaft significantly weakened the bone because of increases in local stresses by the defect (h). Hipp found that a hole that reduced the cross-sectional area of the bone by less than 40% reduced the torsional strength of the bone by 70%. These results suggest that the 50% loss originally stated by Fidler (B) as the cutoff for prophylactic fixation may be an underestimation. Hipp et al (l) also found that the location and shape of endosteal defects affected the degree of strength reduction in the bone which would therefore affect the risk of fracture. If a defect causing a 50% loss in cross sectional area is in the center of the femoral diaphysis, the strength of the bone is reduced by 60%. However, if an identical defect is located such that the thinnest wall was at the point of maximal bending stress, the strength reduction was greater than 90%.
In bones subjected to bending, it is the location of the defect that is important in determining the amount of strength reduction. The length of the defect along the long axis of the bone has a large effect on torsional strength. A long defect with the same decrease in cross sectional area as a small defect will cause a greater reduction in strength than the smaller defect. The length of the defect does not significantly affect bone strength if bones are subjected to bending (5,l). Studies need to be conducted that study the effect of combinations of torsion and bending to determine how they affect the strength of the bone. Biomechanics and computer models promise to improve the accuracy of fracture risk prediction, but these methods are not yet available for everyday use.
In summary, an orthopedic surgeon calculating risk of pathologic fracture is likely to focus most of his or her attention on the appearance of lesion of plain radiographs. The authors recommend that the size of the lesion be considered in the context of the other factors that are mentioned by Mirels et al (1). When the boundaries, or dimensions, of a lesion are uncertain, the threshold for orthopedic stabilization should be lowered.
In certain locations such as the femoral neck, the peritrochanteric region of the femur, and the junction between the humeral head and the humeral metaphysis, the risk and disability from pathologic fracture are so great that orthopedic stabilization should be used in virtually all cases. Only very small, well-delineated lesions in these high-risk locations should be treated non-operatively. If the surgeon chooses non-operative care for a small lesion in a high risk location, careful follow-up is required since the lesions may progress to fracture before the treatment is completed.
