Data from randomized, controlled trials of zoledronic acid were retrospectively analyzed to assess the effect of pathologic fractures on survival in patients with malignant bone disease.
Data from randomized, controlled trials of zoledronic acid were retrospectively analyzed to assess the effect of pathologic fractures on survival in patients with malignant bone disease.
A Cox regression model was used to estimate the effect of fractures (time-dependent variable) on survival in patients with stage III multiple myeloma or bone metastases from solid tumors enrolled in 3 large trials. Patients were randomized to receive zoledronic acid, pamidronate, or placebo every 3–4 weeks for up to 24 months (prostate cancer, breast cancer, and multiple myeloma) or up to 21 months (lung and other solid tumors).
A total of 3049 patients with multiple myeloma (n = 513), breast (n = 1130), prostate (n = 640), or lung cancer or other solid tumors (n = 766) were included in this analysis. Patients with multiple myeloma had the highest fracture incidence (43%), followed by breast (35%), prostate (19%), and lung cancer (17%). In all tumor types except lung, pathologic fracture was associated with a significant increase in risk of death, and breast cancer patients had the greatest increased risk. After adjustment for baseline characteristics, including performance status and prior skeletal complications, breast cancer patients who developed a pathologic fracture on study had a significant 32% increased risk of death relative to patients without a fracture (hazard ratio = 1.32; P < .01); patients with multiple myeloma or prostate cancer had a >20% increased risk of death.
These results suggest that fractures are associated with increased risk of death in patients with malignant bone disease. Therefore, preventing fractures is an important goal of therapy. Cancer 2007. © 2007 American Cancer Society.
Metastasis to bone is common in patients with advanced malignancies, particularly in patients with breast, prostate, lung, kidney, and bladder cancer.1, 2 Bone lesions from multiple myeloma or bone metastases from solid tumors can dramatically increase bone resorption, resulting in skeletal complications such as pathologic fractures, spinal cord compression, hypercalcemia of malignancy, and severe bone pain requiring palliative radiotherapy. These are devastating events for patients that may substantially reduce their functional independence and quality of life and—in some cases—may hasten their death.3–5 Median survival for patients with breast cancer, multiple myeloma, or prostate cancer ranges from approximately 21 to 33 months after initial diagnosis of bone metastases or bone lesions.2 Therefore, although the diagnosis of malignant bone disease generally signifies that the cancer is incurable, patients with bone metastases may survive for an extended period of time at risk for developing skeletal complications. In the placebo arms of several large, randomized, phase 3 trials, most patients experienced at least 1 on-study skeletal-related event (SRE), and pathologic fractures occurred in 39% of patients with breast cancer, in 22% of patients with prostate cancer, and in 22% of patients with bone metastases from lung cancer or other solid tumors during 12, 15, and 21 months of follow-up, respectively.6–8 Consequently, pathologic fractures are a significant clinical concern in these patient populations, and preventing or delaying fractures is an important treatment objective, particularly in patients with advanced breast cancer.
Currently, bisphosphonates are the standard of care for the prevention of skeletal complications in patients with bone metastases. In a systematic review of 8 studies in patients with bone metastases secondary to breast cancer, bisphosphonates were found to reduce the overall risk of skeletal complications by 14% and to reduce the incidence of fractures by 28% to 37%.2, 9 Treatment guidelines from the American Society of Clinical Oncology recommend either intravenous pamidronate (Aredia, Novartis, Basel, Switzerland; 90 mg) or intravenous zoledronic acid (Zometa, Novartis; 4 mg) every 3 to 4 weeks for patients with evidence of bone destruction as a result of bone metastases secondary to breast cancer or stage III multiple myeloma.10, 11 In randomized, placebo-controlled trials, pamidronate significantly reduced the incidence of skeletal complications and prolonged the median time to first skeletal complication in patients with predominantly osteolytic lesions from advanced breast cancer or multiple myeloma.11–14 Subsequently, in a large, randomized, phase 3 comparative trial, zoledronic acid was shown to be superior to pamidronate and provided a significant reduction in the overall risk of skeletal complications beyond that achieved with pamidronate both in the overall study population and, notably, in the subset of patients with breast cancer.15, 16 The efficacy of zoledronic acid in metastatic bone disease has been clearly demonstrated in separate randomized, placebo-controlled trials in patients with breast cancer, multiple myeloma, prostate cancer, and lung cancer or other solid tumors.6, 8, 15, 17
Although correlations between pathologic fractures and survival have been studied in patients with prostate cancer, this correlation has not been investigated specifically in patients with bone metastases either with or without bisphosphonate therapy.18 Therefore, we conducted a retrospective analysis of 3 randomized, controlled trials of zoledronic acid to assess the impact of pathologic fractures on survival in patients with multiple myeloma or bone metastases from solid tumors.
This was a retrospective analysis of 3 randomized, double-blind, controlled phase 3 trials of zoledronic acid in adult patients with bone metastases/lesions from stage IV breast cancer or Durie-Salmon stage III multiple myeloma,15, 19 hormone-refractory prostate cancer,7, 17 or nonsmall-cell lung cancer (NSCLC) or other solid tumors.8, 20 Other eligibility criteria included Eastern Cooperative Oncology Group (ECOG) performance status ≤2, corrected serum calcium ≤8.0 mg/dL, and serum creatinine ≤3.0 mg/dL. Patients with prostate cancer had serum testosterone <50 ng/dL.
Patients in these trials were randomized to treatment with zoledronic acid (4 mg or 8/4 mg) or control (90 mg pamidronate for breast cancer and multiple myeloma and placebo for prostate cancer and NSCLC and other solid tumors) every 3 to 4 weeks for up to 24 months (breast cancer, multiple myeloma, and prostate cancer), or 21 months (NSCLC and other solid tumors).
Skeletal-related events were defined as pathologic fracture (vertebral or nonvertebral), spinal cord compression, surgery to bone, radiation therapy to bone, or a change in antineoplastic therapy to treat bone pain (prostate cancer patients only). Hypercalcemia of malignancy was not included as an SRE in these analyses. The specifications for diagnosing each of these events have been described previously.7, 19, 20 Radionuclide bone scans and plain radiographs of the axial skeleton were performed at baseline and at protocol-specified time intervals on study for central review by a radiologist who was blinded to treatment assignment. Each pathologic fracture was documented on a plain radiograph and evaluated by the central radiologist.
Follow-up for survival could be performed for all patients from the time that they enrolled until the last patient's final visit (conclusion of trial).
Statistical analyses were performed by tumor type on the safety population, which included all randomized patients who received study medication and had at least 1 postbaseline safety evaluation. Survival information was collected for each patient at the time of death or, for patients who remained alive but no longer received study medication, at 6-month intervals until the trial concluded. All available survival data were included in the analyses. A Cox regression model was used to assess the effect of fractures on survival in 2 ways: 1) any fracture or 2) vertebral and nonvertebral fractures. Depending on the type of analysis, the first on-study pathologic fracture or the first on-study vertebral fracture and the first on-study nonvertebral fracture were used in the analysis. Each type of these first events was a time-dependent variable in its respective Cox regression model. Time to death was defined as the time from randomization to the date of death for patients who died, and patients were censored at the date of the last evaluation if they were alive at that time. A study drug treatment group was included in the Cox regression model. The Cox regression model for patients with breast cancer was stratified by current therapy (hormonal or chemotherapy) and the model for patients with NSCLC or other solid tumors was stratified by cancer type. In addition, an adjusted model included the following variables: history of SRE before study entry and baseline ECOG performance status. The 2-sided statistical significance level was 0.05.
Data were available for a total of 3049 patients, including patients with breast cancer (n = 1130), multiple myeloma (n = 513), prostate cancer (n = 640), and lung cancer or other solid tumors (n = 766). Patient characteristics by tumor type and on-study fracture status are summarized in Table 1. Patients with breast cancer were treated with hormonal therapy, chemotherapy, or both. All patients with prostate cancer had progressed while receiving hormonal therapy and were considered hormone refractory. The majority of patients (≈80%) had a baseline ECOG performance status of 0 or 1, and, with the exception of prostate cancer patients, the majority of patients had experienced at least 1 SRE before study entry.
|Breast cancer||Multiple myeloma||Prostate cancer||NSCLC and other solid tumors|
|Pathologic fracture, n = 188||No pathologic fracture, n = 336||Pathologic fracture, n = 205||No pathologic fracture, n = 401||Pathologic fracture. n = 220||No pathologic fracture, n = 293||Pathologic fracture, n = 122||No pathologic fracture, n = 518||Pathologic fracture, n = 127||No pathologic fracture, n = 639|
|Age, y median (range)||56 (26–88)||55 (24–85)||59 (28–88)||59 (31–95)||62 (36–89)||62 (33–87)||74 (45–90)||72 (37–90)||62 (40–86)||63 (25–88)|
|ECOG PS, No. (%)|
|0–1||148 (78.7)||286 (85.1)||165 (80.5)||358 (89.3)||154 (70.0)||231 (78.8)||111 (91.0)||475 (91.7)||110 (86.6)||534 (83.6)|
|≥2||39 (20.7)||50 (14.9)||39 (19.0)||41 (10.2)||65 (29.5)||62 (21.2)||11 (9.0)||42 (8.1)||17 (13.4)||102 (16.0)|
|missing||1 (0.5)||0||1 (0.5)||2 (0.5)||1 (0.5)||0||0||1 (0.2)||0||3 (0.5)|
|Prior SRE, No. (%)|
|Yes||142 (75.5)||172 (51.2)||160 (78.0)||209 (52.1)||193 (87.7)||223 (76.1)||56 (45.9)||158 (30.5)||103 (81.1)||422 (66.0)|
|No||46 (24.5)||164 (48.8)||45 (22.0)||192 (47.9)||27 (12.3)||70 (23.9)||66 (54.1)||360 (69.5)||24 (18.9)||217 (34.0)|
Patients with breast cancer had a median age of 57 years (range, 24–95 years), and baseline characteristics were well balanced between patients with an on-study fracture versus those without an on-study fracture.
Among multiple myeloma patients, the median age was 62 years (range, 33–89 years). Overall, the majority of multiple myeloma patients had an ECOG performance status of 0 or 1. Among patients with an on-study pathologic fracture, significantly more had an ECOG performance status of 0 or 1 versus ≥2 (70% vs 30%; P = .03). In addition, significantly more patients with prior SREs developed pathologic fractures on study (88% vs 12%; P = .001).
The median age was 73 years (range, 37–90 years) in patients with prostate cancer. Overall, the majority of patients had an ECOG performance status of 0 or 1. Compared with other tumor types, patients with prostate cancer experienced fewer SREs before study entry, and significantly more patients without prior SREs developed pathologic fractures on study (54% vs 46%; P = .002).
Similar to the studies in the other tumor types, patients with NSCLC or other solid tumors had a median age of 63 years (range, 25–88 years). Overall, the majority of patients had an ECOG performance status of 0 or 1, and significantly more patients with prior SREs developed pathologic fractures on study (81% vs 19%; P = .001).
For each type of cancer diagnosis the percentage of patients who experienced at least 1 on-study pathologic fracture varied (Table 2). The incidence of pathologic fracture ranged from 17% to 43% and was highest among patients with multiple myeloma (43%) or breast cancer (35%).
|Patients wth fractures,* No. (%)||Follow-up, mo|
|Breast cancer, n = 1130||737 (65)||393 (35)||161 (14)||281 (25)||25|
|Multiple myeloma, n = 513||293 (57)||220 (43)||122 (24)||126 (25)||25|
|Prostate cancer, n = 640||518 (81)||122 (19)||39 (6)||87 (14)||24|
|NSCLC and other solid tumors, n = 766||639 (83)||127 (17)||59 (8)||73 (10)||21|
The percentages of patients who were alive or lost to follow-up versus deceased at study end (treatment duration of approximately 2 years for each cancer type) are shown in Table 3 according to tumor type, treatment group, and on-study fracture status. Although the survival status of patients lost to follow-up is unknown, in general the survival of patients was highest among multiple myeloma patients and lowest among patients with NSCLC or other solid tumors. In patients with metastatic NSCLC, the known mortality rate over the period of 21 months was approximately 80%. In patients with breast cancer, the proportion of patients who had an on-study fracture and were deceased at study end was greater than for patients who did not have an on-study fracture. However, the proportion of patients with multiple myeloma, prostate cancer, or NSCLC or other solid tumors who were deceased at study end was similar in patients with fractures compared with patients without fractures on study.
|Patients, no. (%)|
|Deceased||Alive or lost to follow-up||Deceased||Alive or lost to follow-up|
|Total||217 (55)||176 (45)||350 (47)||387 (53)|
|Zoledronic acid*||142 (56)||111 (44)||226 (46)||262 (54)|
|Pamidronate||75 (54)||65 (46)||124 (50)||125 (50)|
|Total||80 (36)||140 (64)||106 (36)||187 (64)|
|Zoledronic acid*||49 (34)||97 (66)||76 (38)||124 (62)|
|Pamidronate||31 (42)||43 (58)||30 (32)||63 (68)|
|Total||84 (69)||38 (31)||366 (71)||152 (29)|
|Zoledronic acid*||47 (66)||24 (34)||254 (70)||107 (30)|
|Placebo||37 (72)||14 (27)||112 (71)||45 (29)|
|NSCLC and other solid tumors|
|Total||105 (83)||22 (17)||538 (84)||101 (16)|
|Zoledronic acid*||62 (86)||10 (14)||374 (84)||73 (16)|
|Placebo||43 (78)||12 (22)||164 (85)||28 (15)|
Results of the correlative Cox regression analysis using fractures as a time-dependent variable are expressed as hazard ratios of death for patients with a pathologic fracture (developing on study) compared with patients without a pathologic fracture. The hazard ratios by tumor type are shown in Table 4. In this analysis, the most significant correlation between pathologic fractures and death occurred in patients with breast cancer.
|Unadjusted model||Model adjusted for covariates*|
|Hazard ratio (95% CI)||P||Hazard ratio (95% CI)||P|
|Any fracture||1.52 (1.27, 1.81)||<.01||1.32 (1.10, 1.59)||<.01|
|Vertebral fracture||1.42 (1.13, 1.78)||<.01||1.19 (0.94, 1.51)||.15|
|Nonvertebral fracture||1.35 (1.12, 1.64)||<.01||1.24 (1.02, 1.51)||.03|
|Any fracture||1.44 (1.06, 1.95)||.02||1.26 (0.92, 1.72)||.15|
|Vertebral fracture||1.40 (0.99, 1.98)||.06||1.20 (0.84, 1.71)||.31|
|Nonvertebral fracture||1.27 (0.89, 1.79)||.18||1.18 (0.83, 1.67)||.36|
|Any fracture||1.29 (1.01, 1.65)||.04||1.23 (0.96, 1.57)||.10|
|Vertebral fracture||1.29 (0.87, 1.91)||.20||1.11 (0.75, 1.66)||.60|
|Nonvertebral fracture||1.27 (0.96, 1.68)||.09||1.28 (0.97, 1.69)||.08|
|NSCLC and other solid tumors‡|
|Any fracture||1.08 (0.87, 1.34)||.49||1.06 (0.85, 1.32)||.60|
|Vertebral fracture||1.26 (0.94, 1.70)||.13||1.20 (0.89, 1.62)||.24|
|Nonvertebral fracture||0.96 (0.73, 1.28)||.79||0.97 (0.73, 1.29)||.86|
Patients with breast cancer who developed pathologic fractures had a statistically significant 52% increase in the risk of death (not adjusted for baseline characteristics) compared with patients who had no fractures (P < .01). Moreover, after adjusting for previous SREs and baseline ECOG performance status ≥2, pathologic fractures correlated with a statistically significant 32% increase in the risk of death (P < .01) in patients with breast cancer. Similarly, among patients with multiple myeloma or prostate cancer, pathologic fractures were associated with a statistically significant increased risk of death, with hazard ratios of 1.44 (P = .02) and 1.29 (P = .04), respectively (unadjusted for baseline characteristics). After adjusting for prior SREs and baseline ECOG performance status, the risk of death associated with a pathologic fracture was increased by approximately 20% in each of these patient populations, which showed a trend toward statistical significance. In contrast, no significant correlations were observed in patients with NSCLC or other solid tumors.
Results of the Cox regression analysis used both vertebral and nonvertebral fractures as time-dependent variables, and are expressed as hazard ratios of death for patients who developed either of these fractures while on study compared with patients who did not. The hazard ratios by tumor type are shown in Table 4.
The risk of death was significantly associated with both vertebral and nonvertebral fractures in the unadjusted model for breast cancer patients compared with patients who had no such fractures (P < .01 for both). After adjusting for previous SREs and baseline ECOG performance status, nonvertebral fractures maintained a statistically significant correlation with the risk of death (24% increase; P = .03), and vertebral fractures correlated with a trend toward increased risk of death (19% increase; P = .15). No statistically significant correlation was observed between vertebral or nonvertebral fractures and risk of death in patients with multiple myeloma or prostate cancer. However, nonvertebral fractures were associated with a trend toward increased risk of death in patients with prostate cancer (27% increase; P = .09), which was unaffected by adjustment for covariates. In patients with NSCLC or other solid tumors, vertebral fractures were associated with a trend toward increased risk of death, (26% increase; P = .13) in the unadjusted analysis, but the correlation was poorer after adjustment.
Skeletal complications, particularly fractures, are an important healthcare concern in patients with advanced cancer. Evaluation of the placebo arms of randomized, controlled studies has provided valuable insight into the natural history of malignant bone lesions and their associated skeletal complications. Among patients randomized to placebo, pathologic fractures occurred in >30% of breast cancer patients and in approximately 40% of patients with multiple myeloma.6, 21 Fractures may cause severe bone pain, limit mobility, and require surgery and hospitalization for treatment.22 Therefore, fractures can interfere with functional independence and may shorten survival. Evidence from studies in the elderly and patients with chronic kidney disease has demonstrated that osteoporotic fractures are associated with increased mortality.23–26 In fact, hip fractures in the elderly have been shown to reduce life expectancy by 1.8 years compared with an age- and sex-matched general population.23
The large database from the international, multicenter, randomized phase 3 clinical studies of zoledronic acid provided the opportunity to investigate the association between pathologic fractures and survival. The primary objective of these studies was not to examine risk factors for survival; however, the high incidence of pathologic fractures and morbidity in patients with bone lesions from advanced cancer in these studies allows a robust comparison between outcomes in patients based on the occurrence of fractures. These analyses demonstrated a significant increase in risk of death after the occurrence of pathologic fractures. This correlation was observed in patients with multiple myeloma and patients with bone metastases from breast or prostate cancer but not in patients with NSCLC or other solid tumors, presumably because of the short median survival (≈6 months) of this patient population.8 In that group, the risk of death from disease progression may preempt the risk of fractures. Moreover, the majority (69%) of patients with NSCLC or other solid tumors had already experienced at least 1 SRE before study entry, and many of these events might have been pathologic fractures.27 Therefore, the effects of pathologic fractures before study entry could confound on-study correlative analyses for these patients. Indeed, patients with an SRE before study entry had poorer clinical outcomes, including a 41% increase in risk of SREs compared with patients who had no history of SREs before study entry.27 Therefore, fracture reduction may, in part, explain the survival benefit observed in these analyses.
This was an exploratory analysis that did not have the statistical power to formally test whether vertebral or nonvertebral fractures have a greater influence on survival. In patients with breast cancer, the correlation between nonvertebral fractures and survival maintained statistical significance even after adjustment for SRE history and ECOG performance status, whereas significance was not maintained for vertebral fractures. However, there were approximately twice as many patients with nonvertebral fractures compared with vertebral fractures. For multiple myeloma and prostate cancer, there was a consistent modest correlation between either fracture type and increased risk of death, whereas there was no apparent correlation between nonvertebral fractures and survival in patients with NSCLC or other solid tumors.
Although patients with bone metastases consistently have been reported to be at an increased risk for skeletal complications, there are limited data on the correlation between skeletal complications in general, or pathologic fractures in particular, and survival in patients with multiple myeloma and in patients with bone metastases from solid tumors. For most patients, the development of a symptomatic pathologic fracture is a devastating event that heralds the end stage of their disease. Several studies have reported increases in mortality ranging from 23% to 90% among individuals with prevalent vertebral fractures.28 Older individuals who have prevalent vertebral fractures also have an increased risk of mortality. Kado et al.29 reported that women with at least 1 new fracture had an age-adjusted 32% increased risk of mortality compared with women without incident vertebral fractures. Recently, Oefelein et al.18 reported a correlation between pathologic fractures and shorter survival in patients with nonmetastatic prostate cancer. Moreover, of patients who receive surgery for a pathologic fracture, half will die within 6 months of surgery, and there are relatively few long-term survivors.30 Therefore, fractures have now been associated with a significantly increased risk of death in multiple oncology populations. However, the basis for this correlation is not clear. Several factors may be involved; for example, mortality due to surgery for pathologic fractures, loss of mobility and functional independence, increased risk of deep vein thrombosis, and other complications such as clinical features or biology of the disease. Clinical studies may also enroll patient populations with more aggressive bone disease. These factors may have resulted in the differences observed in the present analyses of multiple tumor types. In the study of breast cancer or multiple myeloma, all patients received bisphosphonate treatment; however, the studies in patients with prostate cancer or with NSCLC or other solid tumors were placebo-controlled. Yet the association between pathologic fractures and death remained despite treatment, and was most apparent in breast cancer.
Historically, surgery and radiation therapy have been the mainstays for palliation of pathologic fractures and bone pain. Recently, however, zoledronic acid has demonstrated efficacy in the management of bone pain and prevention of SREs, including pathologic fractures, in a wide range of tumors, including breast cancer, multiple myeloma, prostate cancer, and NSCLC and other solid tumors.31 The mechanism of action of bisphosphonates such as clodronate, pamidronate, and zoledronic acid in preventing bone resorption involves the inhibition of osteoclast-mediated bone resorption, thereby reducing the risk of SREs.32 Several studies have demonstrated long-term efficacy of bisphosphonates in reducing the incidence of skeletal complications in patients with bone metastases.7, 8, 13, 15, 17, 33 In a recent Cochrane Database Review, Pavlakis et al.9 reported that, compared with placebo or no treatment, intravenous bisphosphonates and oral clodronate significantly reduced the risk of skeletal complications in breast cancer patients with bone metastases (P ≤ .05). Of all the bisphosphonates analyzed, zoledronic acid produced a profound reduction of 41% in the risk of skeletal complications. Consistent with these findings, Kohno et al.6 also demonstrated a reduction in the risk of all types of SREs, including pathologic fractures, with zoledronic acid compared with placebo.
On the basis of preclinical and clinical studies, zoledronic acid is an effective inhibitor of bone resorption and has a favorable tolerability profile.8, 15, 17, 34 Several studies suggest that treatment with zoledronic acid should be proactive and initiated as soon as evidence of bone metastases is visible instead of delaying treatment until fractures develop. Once a fracture occurs, the patient is at an increased risk for a variety of complications that can hasten death. Consequently, the goal of therapy is to prevent these devastating events. Zoledronic acid provides a significant benefit to patients with bone metastases by preventing skeletal complications and reducing bone pain. Therefore, it may allow patients to preserve mobility and functional independence and reduce the risk of mortality from the complications associated with fractures.