The epidemiology of cancer patients with proximal femoral lesions and evaluation of surgical decision tools: A prospective cohort study

Abstract

Introduction

This study evaluated the outcomes of orthopaedic oncological surgical decisions in cancer patients with proximal femoral lesions and the accuracy of fracture prediction tools.

Methods

Patients with proximal femur metastatic cancer or multiple myeloma lesions presenting for prophylactic surgical evaluation between March 2018 and July 2023 were recruited with follow-up until March 2026. Femoral fracture risk was evaluated using CT finite element analysis (CTFEA) and Mirels' score. Patients were randomized, enabling 2/3 of their surgeons to receive the results of their CTFEA.

Results

One hundred and ten patients completed the study. The patient 36-month patient survival rate from the time of orthopaedic oncology referral was 44%, with no difference in the survival estimates of patients who underwent prophylactic surgery and patients with no surgery (P = 0.8. Four patients fractured. CTFEA predicted fracture in all and Mirels' score in two. There was no statistically significant difference between the sensitivity (100% vs 50%), specificity (72% vs 70%), positive prediction value (27% vs 13%), and negative prediction value (100% vs 94%) between CTFEA and the Mirels' score. CTFEA advised surgery in 57% (36/63) patients, but in only 27/36, was the clinical decision surgery.

Conclusion

Compared to publications prior to modern therapy, patient survival improved. Having prophylactic proximal femur surgery because of metastasis was not found to affect survival time. While CTFEA correctly identified all four fracture patients, following its prophylactic surgery recommendations would have led to unnecessary surgeries. Orthopaedic oncologists disregarded the CTFEA recommendation to perform surgery in 25% of patients.

1. Introduction

When solid tumors metastasize to bone the cancer is considered to be stage 4 and incurable. Today, with modern therapies, many patients with stage 4 cancers can achieve high-quality longer lives, with some even having long-term remissions (Geijteman et al., 2024). This changing epidemiology affects the decisions of orthopaedic oncologists when they evaluate cancer patients with a proximal femur lesion (Pulte et al., 2019; de Groot et al., 2026).

Between 2010 and 2015, 5.1% of cancer patients in the U.S. SEER Program (Surveillance, Epidemiology, and End Results) database were diagnosed with bone metastases (Ryan et al., 2022). At the time of the initial diagnosis of their cancer, 18.8 per 100,000 individuals had bone metastases. This data does not include cases of multiple myeloma because the osteolytic bone lesions associated with it are caused by plasmacytes, which originally reside in bone marrow and therefore are not technically metastases. Breast, prostate, lung, kidney and thyroid cancers commonly metastasize to the femur (Hershkovich et al., 2023). Specific data about the incidence of femoral or proximal femoral metastases is not available in the SEER database. Based on analysis of whole-body bone scans, 15% of bone metastases were found to involve the femur (Zacharia et al., 2018).

Patients with proximal femoral metastases present to the orthopaedic surgeon either because of associated bone pain or lesions identified on scans done in their cancer surveillance. It is estimated that there are about 250 academic orthopaedic oncologists in the U.S (Mener et al., 2025). Patients with metastatic proximal femur lesions are often referred to them for evaluation. Their decision whether to perform prophylactic surgery is based on the geometric and anatomical specifics of the lesion, an estimation of fracture risk, the patient's response to therapy, clinical status and expected survival. These parameters also affect the surgeon's choice of the surgical technique. The Mirels' score is often used to estimate the fracture risk of metastatic bone lesions (Mirels, 1989).

We conducted a prospective cohort study among patients with metastatic or multiple myeloma proximal femur bone lesions. The study aims were to follow the clinical course of these patients over time and to assess the results of orthopaedic oncological decisions related to their proximal femoral lesions in the era of recent modern therapies. Additionally, we studied whether making CT finite element analysis (CTFEA) of fracture risk available to orthopaedic oncologists would affect and improve their treatment recommendations.

2. Methods

2.1. Participants

Patients presenting with metastatic cancer or multiple myeloma bone disease of the proximal femur to the Israeli National Unit of Orthopaedic Oncology for evaluation between March 2018 and July 2023 were asked to participate in the study. Follow-up of patients was until March 2026. The study received institutional ethics committee approval from the Tel Aviv Sourasky Medical Center (TLV-0532-17) and was registered at ClinicalTrials.gov (NCT03453905). All participants signed informed consent. The inclusion criteria were: Patients aged 25–90 with metastatic or multiple myeloma lesions of the proximal femur with recent CT scans or were willing to undergo a CT scan of their femurs; patients whose life expectancy was estimated to be >2 months. Exclusion criteria were patients with previous lower extremity pathology; history of previous radiation to the lower limb; metal implants in the femur; radiograph showing a cortical fracture line.

Patients were evaluated for fracture risk and the need for prophylactic surgery by orthopaedic oncologists. Initial recommendations were based on clinical data, plain X-rays, Mirels scores and a CT scan if it was available at the time of initial presentation to the orthopaedic oncology unit. CT scans were ordered for any patient who did not have a scan. An automated CTFEA (Simfini-Tumor, PerSimiO, Beersheba, Israel) was performed on all CT scans, using no clinical data other than patient weight. The system requires a CT of at least 2/3 of both femurs with a soft tissue kernel filter. The Simfini-Tumor categorizes results according to: no, low, moderate, high, and very high fracture risk groupings with surgery recommended in the latter two groups. The details of the Simfini-Tumor CTFEA evaluation tool have been previously described (Yosibash et al., 2007; Sternheim et al., 2018). After the initial clinical decision of the orthopaedic oncologist, the orthopaedic oncological team was provided with available CTFEA results according to a computer program which randomized access to results in a 2:1 yes/no ratio. For cases in which CTFEA results were provided, a final treatment decision was then made if prophylactic surgery was needed.

The Mirels' score is based on four characteristics: anatomical site of the lesion; nature of the lesion; size of the lesion; and pain (Mirels, 1989). For each characteristic the grading is scored between 1 and 3. Proximal femoral lesions receive a site score of three. The Mirel's recommendation for a score ≥ 9 is prophylactic surgery.

All study data were recorded for analysis in a separate database. At the end of the study, all hospital computerized information was reviewed to update the study database. At that time other relevant databases were checked when needed to determine if a patient was still living and if deceased, the date of death.

2.2. Statistical analysis

Statistical analysis was performed using the Statistical Analysis System (SAS Institute Inc., Cary, North Carolina, U.S., version 9.4 M8). Independent categorical variables were compared using the Chi-square test. Comparison between the sensitivity, specificity, of CTFEA and Mirels' Score ≥ 9 for fracture prediction was performed by the McNemar test. Positive and negative predictive values of the two methods were compared using the Leisenring, Alonzo, and Pepe test. Survival analysis was performed with PROC LIFETEST censoring survivors at latest follow-up.

3. Results

3.1. Epidemiologic results

One hundred and fifteen patients with metastatic or multiple myeloma proximal femur lesions were recruited into the study and signed informed consent. Approximately 80% of eligible patients were recruited. One hundred and ten patients completed the epidemiology portion of the study with full data of their clinical course. Table 1 presents the demographics of these patients. Fifty-eight percent of patients were female.

Table 1.

Patient demographics.

Variable	Mean + S.D.
Age (years)	64.2 ± 13.1
Weight (kg.)	74.7 ± 16.1
Height (cm.)	167 ± 9.7
BMI	26.9 ± 5.0
Mirels score	8.5 ± 1.2

The primary cancer of the 110 study patients is presented in Table 2.

Table 2.

Primary cancer of study patients.

Cancer type	Number of patients	Frequency %
Breast	35	31.5
Prostate	15	13.6
Multiple myeloma	13	11.8
Lung	11	10.0
Renal cell carcinoma	11	10.0
Unknown origin	5	4.6
Other	20	18.2

Fig. 1 is a Kaplan-Meier survival graph for all subjects according to major cancer types. Overall survival at 36-months was 44%. Survival at 36-months depended on cancer type and ranged from 69% (9/13) for multiple myeloma, 55% (6/11) for renal cell carcinoma, 43% (13/31) for breast, 40% (6/15) for prostate, to 20% (2/10) for lung. Thirty-one percent of patients were still living at the end of their study follow-up. Eighteen out of 67 patients with five years or more of follow-up were still living.

Fig. 1.

Kaplan-Meier survival estimate of 110 study patients according to cancer types.

Fig. 2 charts the clinical course of patients through the study. Forty-five percent of the patients underwent prophylactic surgery. Fig. 3 is a Kaplan-Meier survival estimate comparing patients who had prophylactic surgery and those who did not. There was no statistical significance between the survival estimates of the two groups, p = 0.8.

Fig. 2.

Study flow chart.

Fig. 3.

Kaplan-Meier survival estimate comparing those who underwent prophylactic surgery with those who did not.

Radiotherapy was given to all patients who did not undergo prophylactic surgery. Thirty-eight percent of patients who had prophylactic surgery received post operative radiotherapy.

3.2. CTFEA and its effect on the expert orthopaedist's decision

Fig. 4 is a flow chart which shows details of CTFEA results, availability and use in surgical decisions. Fourteen of the 110 patients did not have a complete CTFEA analysis. In addition to these 14 cases, there was a delay in the orthopaedic oncological team receiving the results. An additional two patients died within two months of having their CT scan. This left 88 CTFEA scans available to assess whether accessing their results affected orthopaedic oncologists' prophylactic surgery decisions. In 63/88 cases (72%), the orthopaedic oncologists were provided with the CTFEA results before they made their final` decision whether or not to perform prophylactic surgery. In 57% (36/63) of these patients, CTFEA advised surgery, but in only 75% (27/36), was the final orthopaedic oncology team decision surgery. Before seeing the CTFEA results, there was a discrepancy between the initial orthopaedic decision and the recommendation of the CTFEA in 15 patients. In only 3/15 (20%), did the orthopaedic oncology team change their final decision according to the CTFEA recommendation.

Fig. 4.

Flow chart of CTFEA results.

In 10/25 (40%) of the cases when the CTFEA was performed, but the results were not given to the orthopaedic oncologists, there was a discrepancy between the recommendation of the CTFEA and the initial expert's recommendation. In three of the cases in which CTFEA recommended surgery, the orthopaedic oncology team's decision was follow-up. The final decision of the orthopaedic team changed in 1/25 cases even though they had no CTFEA results.

3.3. Fractures in those who did not have prophylactic surgery and post-surgical fractures

Three out of the 47 patients (6%) who the orthopaedic oncological team chose not to perform prophylactic surgery sustained femoral fractures within two months of their enrollment into the study. In two of the patients, the primary cancer was breast, and in one, the primary was lung. All had received radiation therapy prior to the fracture. An additional patient scheduled for prophylactic surgery, fractured one day before the scheduled surgery. Of these four patients, one was still living at the end of the study follow-up, 40 months after induction into the study. CTFEA advised surgery in all four patients and Mirels' score advised surgery in two. No study patient whose CTFEA recommendation was not to perform prophylactic surgery subsequently fractured. Two patients who had prophylactic surgery (CTFEA did not recommend surgery) experienced a fracture in the operated femur, nine months and 12 months following surgery.

3.4. Comparison of the accuracy of Mirels' score and CTFEA

Table 3 compares the sensitivity, specificity, and positive and negative predictive values of Mirels scoring with CTFEA as a fracture risk prediction tool.

Table 3.

Sensitivity, specificity, and positive and negative predictive values according to fracture risk prediction tools.

	Fracture Risk Prediction Tool
Statistical parameter	Mirels ≥ 9	CTFEA	P value
Sensitivity (95% CI)	50% (1%–99%)	100% (100%–100%)	0.50⁎
Specificity (95% CI)	70% (56%–83%)	72% (59%–86%)	1.0⁎
Positive prediction value (95% CI)	13% (0.0%–31%)	27% (4%–46%)	0.34⁎⁎
Negative prediction value (95% CI)	94% (85%–100%)	100% (100%–100%)	0.16⁎⁎

^⁎McNemar's Test.

^⁎⁎Leisenring, Alonzo, and Pepe test.

4. Discussion

One of the aims of the present study was to see if the availability of biological and immunological cancer treatments has affected the survival of patients with metastatic proximal femoral lesions. This study began in 2018 at a time when checkpoint inhibitors had just become available. The use of checkpoint inhibitors such as pembrolizumab has greatly increased the survival rate for non-small cell lung cancer, renal cell carcinoma and triple-negative breast cancer patients. Other biological and immunological treatments have also become available for subtypes of breast, prostate, lung, and kidney cancers that commonly metastasize to bone (Hershkovich et al., 2023).

The current study followed patient survival from the time of referral to the orthopaedic oncology unit for evaluation of proximal femoral metastases or multiple myeloma lesions. The overall 36-month survival was 44%. Twenty-seven percent of patients who had a five-year follow-up were living. Survival at 36-months depended on cancer type ranging from 69% (9/13) for multiple myeloma, to only 20% (2/10) for lung cancer. Hershkovich et al. (Hershkovich et al., 2023) reported the survival of a retrospective cohort of patients with femoral metastases who underwent surgery for impending fracture or fracture from 2004 to 2015. Follow-up was until 2016. The cumulative survival at 36-months, extrapolated from their Kaplan-Meier survival plot, was less than 18%. They found prostate cancer had the best 36-month survival rate and lung cancer the worst. Mehnert et al. (Mehnert et al., 2024) performed a retrospective survival analysis of patients with proximal femoral metastatic or hematopoietic lesions treated during a later period (2014–2020). The cumulative survival at 36-months for their cohort extrapolated from their Kaplan-Meier survival plot was less than 38%. Comparing the present study with these earlier studies it can be concluded that recently developed modalities of treatment have increased the survival of cancer patients even at the stage when they have proximal femoral metastatic or hematopoietic lesions.

In this study, there was no difference between the Kaplan-Meier survival estimates of patients who underwent prophylactic surgery and those who did not undergo prophylactic surgical treatment for their proximal femoral lesions. No corresponding data are available from the Hershkovich et al. (Pulte et al., 2019) study, but they found that patients who underwent prophylactic surgery for femoral metastasis had longer survival rates than those who fractured at their femoral metastatic lesions. Similar findings were reported in a retrospective study based on Veterans Administration data, but their analysis excluded patients who underwent arthroplasty (Philipp et al., 2020).

The decision whether the presence of a proximal femoral metastatic or multiple myeloma lesion should undergo prophylactic surgery to prevent fracture is multifactorial. First is an assessment of the immediate danger of fracture. Second is whether the progression of the lesion can be halted by radiotherapy or oncological therapy. Third is the projected life span of the patients and fourth is whether pain associated with the lesion prevents or limits ambulation. All study patients who did not undergo surgery received radiotherapy. Thirty eight percent of patients who underwent prophylactic surgery received post-operative radiotherapy. Three patients who did not undergo prophylactic surgery fractured within two months after study enrollment. All three received radiotherapy prior to the fracture. Additionally, a fourth patient who was scheduled for prophylactic surgery fractured one day prior to the scheduled operation.

The true accuracy of the CTFEA and Mirels scoring in this study cannot be calculated because the natural history of patients who underwent prophylactic surgery is not known. Accuracy calculations are therefore based only on patients who did not undergo prophylactic surgery. Prior to this study, and in this study, the Orthopaedic Oncology Unit routinely calculated patient Mirels' scores and the orthopaedic team used them as a decision-making aid. A Mirels' score ≥ 9 is generally considered to be an indication for prophylactic surgery (Hershkovich et al., 2023).^. Mirels' developed his scoring system based on a retrospective analysis of 78 metastatic long bone lesions (Mirels, 1989). The orthopaedic oncology team's initial decision did not follow the Mirels' score convention in 26% of the patients. In this study, a Mirels' score of ≥9 was found to have a 50% sensitivity, 70% specificity, 13% positive predictive value and 94% negative predictive value. In a prospective study of femoral metastatic lesions, sponsored by the Musculoskeletal Tumor Society, a Mirel's score of ≥9 was found to have 66.7% sensitivity, 47.9% specificity, 9.8% positive predictive value and 94.4% negative predictive value (Keyak et al., 2005b). That study also compared Mirels' scoring with another fracture risk measurement, CT-Based Rigidity Analysis (CTRA). It found CTRA to have higher accuracy values than Mirels' scoring with 100% sensitivity, 60.6% specificity, 17.6% positive predictive value and 100% negative predictive value, but the difference between the two assessments was not statistically significant. These values are for femoral metastases in general, with 32% of lesions in their study cohort not in the proximal femur. One inherent limitation of CTRA is that it is based on beam theory, which assumes that bones are straight. This makes CTRA analysis less accurate than CTFEA for the proximal femur, which is a curved bone segment, and most suitable for analysis of diaphyseal lesions (Eggermont et al., 2018). The CTFEA used in this study is fully automatic, while CTRA is a manual procedure that needs to be performed by an experienced biomedical engineer. Additionally, the CTRA can be performed only if one femur is tumor-free and it does not consider the weight of the patient and physiological loads.

The accuracy values for CTFEA of proximal femoral lesions in this study are sensitivity 100%, specificity 72%, positive predictive value 27%, and negative predictive value 100%. None were statistically significantly better than the Mirels' score. Because only four fractures occurred, the study was underpowered to show statistical significance between the two methods. Future research should focus on ways to improve CTFEA accuracy and study its value to orthopaedists who are not orthopaedic oncology specialists and who are geographically distant from major medical centers.

Keyak et al. (Keyak et al., 2005a) were pioneers in developing CTFEA methods for predicting proximal femoral strength. Using four-point bending loading conditions of cadaver femora with metastases and with simulated lesions created with burrs, they found CTFEA to be strongly predictive of the measured fracture loads (range: 0.92–0.98) (Damron et al., 2016). Eggermont et al. (Eggermont et al., 2018) prospectively studied whether CTFEA was better in fracture prediction in patients with femoral metastases than experienced clinicians. Their cohort was composed of patients who were referred for treatment at three radiotherapy institutes. Twenty to 25% of eligible patients participated in the study as opposed to approximately 80% in the present study. CTFEA was performed manually and simulated loading was made by a cup on the head of the femur in the axial direction, as opposed to the automated calculations with loading at anatomic points in the present study. They concluded that CTFEA was more accurate in identifying patients with a high fracture risk than experienced clinicians.

A surprising finding of this study was that making CTFEA fracture risk assessment available to the orthopaedic oncologist team had limited impact on their surgical decisions. They chose to perform surgery in 27 out of 36 patients in which CTFEA predicted a high fracture risk and for only three patients they changed their surgical decision according to the CTFEA. The orthopaedic oncologists in this study were from a national orthopaedic oncology center and experienced clinicians. They worked as a team and had an established decision-making process The limited impact of CTFEA may be because they were confident in their usual decision-making process which included many patient specific dynamic variables. This confidence was not unwarranted, since only 3/47 patients in whom they elected not to perform prophylactic surgery subsequently fractured. Their reliance on clinical experience is also reflected in their initial decision, which differed from the Mirels' score recommendations which they had, in 26% of the cases.

The present study has two parts. First is the epidemiological part, which followed the clinical course over time of cancer patients referred to an orthopaedic oncological unit for proximal femur lesions and the need for prophylactic surgery. This part of the study is prospective with a high recruitment rate of eligible patients, a low dropout rate and patient follow-up continued almost three years after the end of recruitment. Radiotherapy treatment was not uniform in the study cohort. While all patients who did not undergo prophylactic surgery received radiotherapy treatment, only 38% of those who underwent prophylactic surgery received radiotherapy as part of their treatment. The study differs from current practice in which radiological assessments are often done by plain radiographs and not using CT scans as in the present study. Additionally, the study patients were from a national orthopaedic oncology unit in a small country without the patient diversity found in other countries. In the second part of the study an orthopaedic oncological unit's decisions of whether patients with proximal femoral lesions required prophylactic surgery were evaluated and compared to Mirels' score and CTFEA. The natural history of the metastatic lesions could only be followed in patients who did not undergo prophylactic surgery. The accuracy of Mirels' score and CTFEA is based on this sub-population alone, and the true accuracy cannot be known. Additionally, the low number of positive events (four fractures) in the study makes for small-sample bias and statistical fragility. Unlike Mirels' score which can be easily calculated by a clinician, the CTFEA used in the study requires special software that is not generally available. Patient weight is needed for calculations and in 11% of cases, because of technical problems, insufficient quality CT scans or the presence of a femoral implant, fracture risk calculations could not be made. The system's parameters may also be different than those of other CTFEAs.

5. Conclusion

In conclusion, the overall 36-month survival of patients in this study from the time they were referred to the orthopaedic oncology unit for evaluation of a proximal femoral lesion was 44%. Survival was dependent on the primary cancer type. Compared to series published prior to the wide availability of modern treatments such as immunotherapy, antibody drug conjugates and CAR T-cell therapy, survival improved. In the present study no difference in survival time was found between those who underwent prophylactic surgery and those who did not. While the CTFEA used in this study had higher specificity, sensitivity, positive and negative predictive values than a Mirels' score ≥ 9 for fracture in patients with proximal femoral metastatic and multiple myeloma lesions, the differences were not statistically significant. CTFEA is usually performed manually by an expert in a time-consuming process. The automated CTFEA system used in this study bypasses this need, but in 11% of patients it could not be performed because of technical problems or the presence of a femoral implant. Making CTFEA fracture risk calculations available to the orthopaedic oncologists in the study had limited impact on their final clinical decisions. While CTFEA correctly identified all four fracture patients in this study and the Mirels' score only two, following CTFEA prophylactic surgery recommendations would have led to unnecessary surgeries.

CRediT authorship contribution statement

Amir Sternheim: Writing – review & editing, Project administration, Methodology, Conceptualization. Zohar Yosibash: Writing – review & editing, Supervision, Resources, Methodology, Data curation, Conceptualization. Charles Milgrom: Writing – review & editing, Writing – original draft, Formal analysis, Conceptualization. Nir Trabelsi: Writing – review & editing, Software, Methodology. Gal Ariel: Data curation. Omri Merose: Data curation. Yair Gortzak: Data curation. Aharon S. Finestone: Writing – review & editing, Methodology, Formal analysis.

Informed consent and patient details

Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.

Studies in human

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place.

This study was approved by the Tel Aviv Sourasky Medical Center.

(Approval No. TLV-0532-17)

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.