Clinically Important Reductions in Physical Function and Quality of Life in Adults with Tumor Prostheses in the Hip and Knee: A Cross-sectional Study

Abstract

Background

Patients with a bone sarcoma who undergo limb-sparing surgery and reconstruction with a tumor prosthesis in the lower extremity have been shown to have reduced self-reported physical function and quality of life (QoL). To provide patients facing these operations with better expectations of future physical function and to better evaluate and improve upon postoperative interventions, data from objectively measured physical function have been suggested.

Questions/purposes

We sought to explore different aspects of physical function, using the International Classification of Functioning, Disability, and Health (ICF) as a framework, by asking: (1) What are the differences between patients 2 to 12 years after a bone resection and reconstruction surgery of the hip and knee following resection of a bone sarcoma or giant cell tumor of bone and age-matched controls without walking limitations in ICF body functions (ROM, muscle strength, pain), ICF activity and participation (walking, getting up from a chair, daily tasks), and QoL? (2) Within the patient group, do ICF body functions and ICF activity and participation outcome scores correlate with QoL?

Methods

Between 2006 and 2016, we treated 72 patients for bone sarcoma or giant cell tumor of bone resulting in bone resection and reconstruction with a tumor prosthesis of the hip or knee. At the timepoint for inclusion, 47 patients were alive. Of those, 6% (3 of 47) had undergone amputation in the lower limb and were excluded. A further 32% (14 of 44) were excluded because of being younger than 18 years of age, pregnant, having long transportation, palliative care, or declining participation, leaving 68% (30 of 44) for analysis. Thus, 30 patients and 30 controls with a mean age of 51 ± 18 years and 52 ± 17 years, respectively, were included in this cross-sectional study. Included patients had been treated with either a proximal femoral (40% [12 of 30]), distal femoral (47% [14 of 30]), or proximal tibia (13% [4 of 30]) reconstruction. The patients were assessed 2 to 12 years (mean 7 ± 3 years) after the resection-reconstruction. The controls were matched on gender and age (± 4 years) and included if they considered their walking capacity to be normal and had no pain in the lower extremity. Included outcome measures were: passive ROM of hip flexion, extension, and abduction and knee flexion and extension; isometric muscle strength of knee flexion, knee extension and hip abduction using a hand-held dynamometer; pain intensity (numeric rating scale; NRS) and distribution (pain drawing); the 6-minute walk test (6MWT); the 30-second chair-stand test (CST); the Toronto Extremity Salvage Score (TESS), and the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30). The TESS and the EORTC QLQ-C30 were normalized to 0 to 100 points. Higher scoring represents better status for TESS and EORTC global health and physical functioning scales. Minimum clinically important difference for muscle strength is 20% to 25%, NRS 2 points, 6MWT 14 to 31 meters, CST 2 repetitions, TESS 12 to 15 points, and EORTC QLQ-C30 5 to 20 points.

Results

Compared with controls, the patients had less knee extension and hip abduction strength in both the surgical and nonsurgical limbs and regardless of reconstruction site. Mean knee extension strength in patients versus controls were: surgical limb 0.9 ± 0.5 Nm/kg versus 2.1 ± 0.6 Nm/kg (mean difference -1.3 Nm/kg [95% CI -1.5 to -1.0]; p < 0.001) and nonsurgical limb 1.7 ± 0.6 Nm/kg versus 2.2 ± 0.6 Nm/kg (mean difference -0.5 Nm/kg [95% CI -0.8 to -0.2]; p = 0.003). Mean hip abduction strength in patients versus controls were: surgical limb 1.1 ± 0.4 Nm/kg versus 1.9 ± 0.5 Nm/kg (mean difference -0.7 Nm/kg [95% CI -1.0 to -0.5]; p < 0.001) and nonsurgical limb 1.5 ± 0.4 Nm/kg versus 1.9 ± 0.5 Nm/kg (-0.4 Nm/kg [95% CI -0.6 to -0.2]; p = 0.001). Mean hip flexion ROM in patients with proximal femoral reconstructions was 113° ± 18° compared with controls 130° ± 11° (mean difference -17°; p = 0.006). Mean knee flexion ROM in patients with distal femoral reconstructions was 113° ± 29° compared with patients in the control group 146° ± 9° (mean difference -34°; p = 0.002). Eighty-seven percent (26 of 30) of the patients reported pain, predominantly in the knee, anterior thigh, and gluteal area. The patients showed poorer walking and chair-stand capacity and had lower TESS scores than patients in the control group. Mean 6MWT was 499 ± 100 meters versus 607 ± 68 meters (mean difference -108 meters; p < 0.001), mean CST was 12 ± 5 repetitions versus 18 ± 5 repetitions (mean difference -7 repetitions; p < 0.001), and median (interquartile range) TESS score was 78 (21) points versus 100 (10) points (p < 0.001) in patients and controls, respectively. Higher pain scores correlated to lower physical functioning of the EORTC QLQ-C30 (Rho -0.40 to -0.54; all p values < 0.05). Less muscle strength in knee extension, knee flexion, and hip abduction correlated to lower physical functioning of the EORTC QLQ-C30 (Rho 0.40 to 0.51; all p values < 0.05).

Conclusion

This patient group demonstrated clinically important muscle weaknesses not only in resected muscles but also in the contralateral limb. Many patients reported pain, and they showed reductions in walking and chair-stand capacity comparable to elderly people. The results are relevant for information before surgery, and assessments of objective physical function are advisable in postoperative monitoring. Prospective studies evaluating the course of physical function and which include assessments of objectively measured physical function are warranted. Studies following this patient group with repetitive measures over about 5 years could provide information about the course of physical function, enable comparisons with population norms, and lead to better-designed, targeted, and timely postoperative interventions.

Level of Evidence

Level III, therapeutic study.

Introduction

Standard treatment after diagnosis of a primary bone sarcoma is usually limb-sparing surgery often combined with neoadjuvant and adjuvant chemotherapy [21]. The operative procedure is extensive, with resection of the affected bone, muscle, and adjacent joint followed by reconstruction with either tumor prostheses, allografts, autografts, or a bone graft prosthetic composite. The muscles may be partly attached to one another and partly to the prosthesis but only seldom can those muscles integrate with it. With the loss of natural bone insertion of surrounding muscles, the biomechanics of the joint and muscles change substantially. Because the patient group is often fairly young at diagnosis, and given that both 5- and 10-year survival rates have been reported to be around 54% to 62% [2, 53, 60], the patients may live many years with the tumor prosthesis. Considering the extensive surgical procedure, one may assume that there is a high risk of reduced physical function over time.

Physical Function

Prior studies describing physical function in this patient group, however, are often limited to the disease-specific Toronto Extremity Salvage Score (TESS) and the Musculoskeletal Tumor Society Score (MSTS). When compared with the International Classification of Functioning, Disability, and Health (ICF), the TESS and MSTS measure some aspects of physical function, but they do not provide a complete clinical picture [19, 46, 61]. It has been suggested that objective measures of physical function such as gait, balance, physical activity, ROM, and muscle strength should serve as important supplements [19, 39, 40]. Patients going through limb-sparing surgery are obliged to know what short- and long-term postoperative physical status they can expect. Combining results from self-reported and objectively measured physical function might provide a more complete and tangible image of expectations. Moreover, objectively measured physical function is often of generic nature, enabling comparisons between different treatment modalities, techniques, and samples, which is important in research and clinical settings. Muscle strength is one example of a generic physical function measure often used in musculoskeletal research. Five studies assessed muscle strength in patients who underwent limb-sparing surgery due to bone sarcoma [4, 7, 8, 12, 16]. One of these studies was a case report of four patients [4], two of whom used manual muscle testing (that is, they did not objectively measure muscle strength) [7, 16], compromising the quality of results. Two studies used dynamometers to measure muscle strength and showed less knee extension, knee flexion, and hip abduction peak torques compared with the nonsurgical limb, but they did not include comparisons to a reference group [9, 12]. Although ROM, pain, balance, and walking capacity are other important generic outcomes of physical function, the prior reports are sparse and comparisons to reference norms are almost nonexistent [19].

Physical Function in Relation to Quality of Life

Quality of life (QoL) is an outcome that is important to patients, and improved QoL often serves as a long-term goal in medicine and rehabilitation [18, 31]. Both children and adults with bone sarcomas have reported inferior QoL compared with population norms [15, 47, 54]. To better target exercise interventions to improve QoL, it is valuable to know what physical function components influence QoL. One study found that greater knee ROM correlated with faster stair climbing, walking speed, and a higher QoL score in 10- to 26-year-old patients [40]. This would indicate that ROM could be an important element to include in exercise interventions. Likewise, other components of physical function (such as pain, muscle strength, walking capacity, and daily tasks) could be correlated with QoL and thereby inform us about potentially important elements for the exercise intervention. In summary, the knowledge about different impairments and their severity, as well as the relationship to QoL, is limited in this patient group.

Research Questions

In this study we sought to explore different aspects of physical function, using the ICF as a framework, by asking: (1) What are the differences between patients 2 to 12 years after a bone resection and reconstruction surgery of the hip and knee following resection of a bone sarcoma or giant cell tumor of bone and age-matched controls without walking limitation in ICF body functions (ROM, muscle strength, pain), ICF activity and participation (walking, getting up from a chair, daily tasks), and QoL? (2) Within the patient group, do ICF body functions and ICF activity and participation outcome scores correlate with QoL?

Patients and Methods

Study Design, Setting

The patients were compared with matched controls in a cross-sectional study. Recruitment and assessments took place between September 2018 and October 2019 in the Musculoskeletal Tumor Section of University Hospital Rigshospitalet in Denmark. Patient recruitment was administered by a physical therapist (LF), an orthopedic resident (CEH), and a consultant orthopaedic surgeon (AV). Recruitment of controls was administered by one author (LF) and four physical therapy students.

Participants

Patients were enrolled if they had been diagnosed with bone sarcoma or giant cell tumor of bone resulting in bone resection and reconstruction with a tumor prosthesis of the hip or knee at the Musculoskeletal Tumor Section, University Hospital Rigshospitalet, between January 2006 and December 2016 (n = 72). At start of recruitment in September 2018, 47 of the initial 72 patients were alive and thus eligible for physical evaluation (Fig. 1). Exclusion criteria were (1) age younger than 18 years, (2) amputation in the lower limb, (3) less than 2 years since initial surgery, (4) pregnancy, and (5) not understanding Danish. Three patients were excluded based on hospital records of amputation. The remaining 44 patients were informed about the study by phone or during outpatient visits at the hospital. A total of 68% (30 patients) were included (Fig. 1). Excluded patients were younger than 18 years (n = 1), pregnant (n = 1), did not understand Danish (n = 1), declined participation for various reasons (n = 7), or did not return calls (n = 4) (Fig. 1).

Fig. 1.

Flowchart of enrollment of patients.

Patients in the control group were recruited from postings at the hospital, university college, and social media. We matched patients in the control group on age ± 4 years and gender, as the two variables have a strong relationship to performance-based tests [24]. We applied a matching factor of 1:1. To reach 30 matched controls, a total of 53 patients in the control group were considered. Exclusion criteria were current and/or long-lasting pain in the lower limb, known diseases that interfered with walking capacity or stair walking, and the use of a walking aid. Although not a reason for exclusion, none of the patients in the control group had had a hip or knee replacement.

Participant Demographics

Included patients had a mean age of 51 ± 18 years and controls 52 ± 17 years (Table 1). The patients were seen 2 to 12 years (mean 7 ± 3 years) after initial surgery. Tumor subtypes were 47% (14 of 30) chondrosarcoma and 30% (9 of 30) osteosarcoma (Table 1). Forty percent (12 of 30) underwent proximal femoral, 47% (14 of 30) distal femoral, and 13% (4 of 30) proximal tibia resection and reconstruction surgery. Mean bone resection length for the proximal femur was 15 ± 3 cm, for the distal femur it was 15 ± 4 cm, and for the proximal tibia it was 14 ± 3 cm. Of the tumor prostheses used, 67% (20 of 30) were the GMRS (Stryker), 27% (8 of 30) were the Segmental (Zimmer Biomet), and 7% (2 of 30) were the Mega C (Link), depending in part on surgeon choice and the prostheses that were generally being used at the time of the resection. Twenty percent (6 of 30) of patients underwent revision, and 3% (1 of 30) had two revisions. Forty percent (12 of 30) had received chemotherapy (neoadjuvant and adjuvant [n = 11] and only neoadjuvant [n = 1]), and 3% (1 of 30) underwent postoperative radiation therapy. In general, chemotherapy (neoadjuvant and adjuvant) was used for the treatment of osteosarcoma and Ewing sarcoma and postoperative external radiation therapy was used in the treatment of Ewing sarcoma in case of poor chemotherapy response and/or only marginal tumor removal. Three patients used crutches daily and four used them from time to time.

Table 1.

Participant demographics

Characteristics		Patients (n = 30)	Controls (n = 30)	p value
Men		53 (16)	53 (16)	0.99
Age in years		51 ± 18	52 ± 17	0.61
BMI in kg/m2		26 ± 4	26 ± 4	0.94
Comorbidity, % yes (n)		43 (13)	33 (10)	0.43
Histology
Chondrosarcoma		47 (14)	N/A
Osteosarcoma		30 (9)
Giant cell tumor		10 (3)
Other		13 (4)

Data are presented as % (n) or mean ± SD; N/A = not applicable.

Description of Experiment, Procedure

All participants were assessed by a physical therapist who was not involved in their initial surgical or postoperative care (LF), with assistance of students, at the Musculoskeletal Tumor Section, University Hospital Rigshospitalet. The questionnaires had been sent to the participants beforehand to be filled out the same day as the appointment or one day ahead. The test session lasted one hour, and tests were carried out in the following order: 6-minute walk test (6MWT), 30 second chair-stand test (CST), ROM, isometric muscle strength, and body weight. Body weight was recorded on a scale (MT32, n^o 0299, BISCO Vægte A/S). One patient declined the test session at the hospital because of long distance but filled in and returned the questionnaires. No adverse effects were seen from the tests.

Variables, Outcome Measures, Data Sources, and Bias

It was not possible to blind the test personnel to the reconstruction site or whether the participant was a patient or a control. To minimize observer bias, standardized test procedures were applied and information such as prosthesis used, time from surgery, whether the patient had undergone any revision surgery, or postoperative care was kept from test personnel. We used the ICF as a framework to list the outcome measures of physical function into the two domains: (1) body function and (2) activity and participation. Developed by the World Health Organization, the ICF is used to standardize and code functioning and disability associated with health conditions [61]. The ICD-10 and the ICF are complementary; together, they provide information on diagnosis plus functioning, giving a broader and more meaningful picture of the health of populations.

ICF Body Function

We measured ROM using a full-circle 1°-increment plastic goniometer with a moveable arm. Hip extension, flexion, and abduction were assessed bilaterally in patients with proximal femur reconstruction and their matched individual in the control group. Knee extension and flexion were assessed bilaterally in patients with distal femur or proximal tibia reconstructions and their matched individual in the control group. All motion directions, except for hip extension, were measured with the patient in a supine position, with the opposite thigh fixed in neutral position [29, 30]. The nonsurgical limb was assessed first.

We tested maximum voluntary isometric muscle strength of (1) knee extension, (2) hip abduction, and (3) knee flexion bilaterally using a hand-held dynamometer (Hoggan microFET2, Hoggan Scientific LLC). The nonsurgical limb was tested first. The tests were performed using a standardized protocol for test positions, fixation, and instructions during the test [41]. We recorded three maximal intensity trials. Data was normalized to body weight (Nm/kg), and percent muscle deficits were expressed as relative to reference $((1-\left(\frac{target value}{reference value}\right))\ast 100)$. Minimum clinical important differences (MCIDs) for isometric knee extension and hip abduction strength were estimated between 20% to 25% [41, 42, 49].

We assessed pain intensity with the commonly used 11-point numeric rating scale (NRS), where 0 equaled “no pain” and 10 equaled “worst pain imaginable” [26]. Three NRS ratings were applied: (1) current pain, (2) worst pain during the past week, and (3) mean pain during the past week. MCID for the NRS pain scale has been reported at 2 points [23].

A pain drawing was used to illustrate pain distribution. The participants were asked to mark the area where they felt pain on a paper silhouette illustration of the human body, including anterior and posterior views. A transparent template with 42 body regions was placed on the pain drawing [34, 37]. Any mark within a body region was scored as “pain present.”

ICF Activity and Participation

Walking capacity was assessed using the 6MWT [3]. The participants were instructed to walk back and forth as quickly as possible for 6 minutes on a 20-meter walking track. The 6MWT has been widely used in numerous patient groups [17]. MCIDs for the 6MWT have been reported to be between 14 and 31 meters [10].

The CST aims to record functional muscle strength and power by evaluating the activity “sit-to-stand” in both younger and older populations [22, 45, 57]. It is scored by counting the maximum amount of complete chair-stand movements during 30 seconds. MCID for the 30-second CST has been estimated to two repetitions [14].

The TESS is a disease-specific self-administered questionnaire measuring physical function [13, 51]. It contains 30 questions answered on 0 to 5 Likert scale. A standardized summary score is calculated ranging from 0 to 100, where higher scores indicate better physical function [13]. In a Danish sample, the TESS showed lower and upper limits of agreement of -12 and 15 points, respectively [51].

Compound Outcome Measure

The European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30) is a QoL questionnaire developed to cover content relevant and important to cancer patients with items pertaining to both ICF body function and activity and participation [35, 50]. It consists of 30 items answered by the patient on Likert scales ranging from 1 to 4 or 1 to 7. Subscale scores for general health, functional scales, and symptom scales were calculated using a scoring algorithm [1]. MCIDs for improvement and deterioration for the functional scales have been estimated within the range of 5 to 20 points [48]. For the correlational analyses, we selected the subscales for general health and physical functioning. Our rationale for this was that general health represented an overall score of QoL and physical functioning represented the same construct as the included outcome measures of physical function described above. Moderate-to-high correlations were expected for physical functioning and low-to-moderate for general health [38, 56].

Ethical Approval

Ethical approval for this study was obtained from the Danish Data Protection Agency (VD-2018-20, 6594) and the Capital Regional Committee on Health Research Ethics (H-18032141).

Statistical Analysis, Study Size

Data were analyzed in SPSS version 25 (SPSS Inc). Results were regarded as statistically significant at p < 0.05. Normality of variables was tested by visual inspection (histograms, probability plots) and the Shapiro-Wilk test. For normally distributed data, the paired t-test was used to evaluate differences between groups along with presentation of mean differences (95% confidence interval). For nonnormally distributed data, between-group differences were analyzed using the Wilcoxon signed rank test. To compare muscle strength data to a previously published study, we performed a subgroup analysis based on reconstruction site [8], where paired t-tests were used despite violating the assumption of normal distribution. We used the Spearman rank correlation coefficient (Rho) to analyze correlations. Because a substantial number of various parameters were examined and tested statistically without any predetermined primary outcome measure (and with no sample size calculations performed), the results of this study cannot be considered confirmatory but are of an exploratory nature. This is also the reason why we did not perform any correction for the multiple testing. To justify pooling of patients with different age and time since surgery, correlations between TESS scores and age (r = -0.07; p = 0.70) and time since surgery (r = 0.25; p = 0.18) were performed. In addition, no differences in TESS scores depending on the three sites of surgery were found using the Kruskal-Wallis test (p = 0.73). One patient declined assessments at the hospital (ROM, muscle strength, 6MWT, CST) but filled in the self-reported outcomes. Missing observations from this patient were not replaced.

Results

Differences Between Patients and Controls

ICF Body Function

The patients showed less hip and knee flexion ROM but similar hip and knee extension ROM compared with controls (Table 2). Mean hip flexion ROM in patients with proximal femoral reconstructions was 113° ± 18° and in controls was 130° ± 11° (mean difference -17° [95% CI -29° to -6°]; p = 0.006). Mean knee flexion ROM in patients with distal femoral reconstructions was 113° ± 29° and in controls was 146° ± 9° (mean difference -34° [95% CI -53° to -14°]; p = 0.002).

Table 2.

Hip and knee passive ROM (in degrees) in the patient and control groups

		Patient group	Control group			Patient group	Control group
ROM by reconstruction site		Surgical limb	Matched to surgical limb	Between-group mean difference (95% CI)	p value	Nonsurgical limb	Matched to nonsurgical limb	Between-group mean difference (95% CI)	p value
Proximal femur
Hip	Flexion	113 ± 18	130 ± 11	-17 (-29 to -6)	0.006	131 ± 9	131± 11	0 (-10 to 11)	0.98
	Extension	8 ± 5	15 ± 8	-8 (-13 to -2)	0.01	9 ± 7	14 ± 7	-6 (-11 to 0)	0.047
	Abduction	34 ± 10	38 ± 5	-4 (-11 to 3)	0.24	33 ± 9	41 ± 7	-8 (-15 to -1)	0.03
Distal femur
Knee	Flexion	113 ± 29	146 ± 9	-34 (-53 to -14)	0.002	140 ± 9	146 ± 9	-6 (-14 to 2)	0.10
	Extension	3 ± 4	3 ± 4	0 (-3 to 3)	0.86	4 ± 3	3 ± 3	0 (-2 to 3)	0.71
Proximal tibia
Knee	Flexion	108 ± 14	151 ± 2	-43 (-66 to -20)	0.009	141 ± 8	151 ± 3	-10 (-22 to 2)	0.07
	Extension	1 ± 2	4 ± 3	-3 (-7 to 0)	0.06	3 ± 5	4 ± 3	-2 (-6 to 3)	0.32

Data are presented as mean ± SD and mean difference (95% CI); hip ROM in patients with proximal femoral reconstructions (n = 11/group); knee ROM in patients with distal femoral (n = 14/group) and proximal tibia (n = 4/group) reconstructions.

Compared with controls, the patients had less knee extension and hip abduction strength in both the surgical and nonsurgical limbs. Mean knee extension strength in the patients’ surgical limb was 0.9 ± 0.5 Nm/kg versus those in the control group 2.1 ± 0.6 Nm/kg (mean difference -1.3 Nm/kg [95% CI -1.5 to -1.0]; p < 0.001) and in patients’ nonsurgical limb it was 1.7 ± 0.6 Nm/kg versus the control group 2.2 ± 0.6 Nm/kg (mean difference -0.5 Nm/kg [95% CI -0.8 to -0.2]; p = 0.003) (Table 3). Mean hip abduction strength in the patients’ surgical limb was 1.1 ± 0.4 Nm/kg versus the control group 1.9 ± 0.5 Nm/kg (mean difference -0.7 Nm/kg [95% CI -1.0 to -0.5]; p < 0.001) and in patients’ nonsurgical limb it was 1.5 ± 0.4 Nm/kg versus the control group 1.9 ± 0.5 Nm/kg (-0.4 Nm/kg [95% CI -0.6 to -0.2]; p = 0.001). Using the limbs of the control group as reference values, muscle strength deficits in the surgical limbs ranged from 28% to 66% (Fig. 2A-C). Muscle strength in patients’ nonsurgical limbs ranged from no deficit to 27% deficit (Fig. 2A-C).

Table 3.

Isometric muscle strength (in Nm/kg) in patients (n = 29) and controls (n = 29)^a

	Patients	Controls	Between-group mean difference (95% CI)	p value	% deficit	Patients	Controls	Between-group mean difference (95% CI)
Muscle group	Surgical limb	Matched to surgical limb				Nonsurgical limb	Matched to nonsurgical limb		p value	% deficit
Knee extension	0.9 ± 0.5	2.1 ± 0.6	-1.3 (-1.5 to -1.0)	< 0.001	57%	1.7 ± 0.6	2.2 ± 0.6	-0.5 (-0.8 to -0.2)	0.003	34%
Knee flexion	0.7 ± 0.2	1.1 ± 0.3	-0.5 (-0.6 to -0.4)	< 0.001	42%	1.0 ± 0.4	1.1 ± 0.3	-0.12 (-0.3 to 0.1)	0.16	11%
Hip abduction	1.1 ± 0.4	1.9 ± 0.5	-0.7 (-1.0 to -0.5)	< 0.001	40%	1.5 ± 0.4	1.9 ± 0.5	-0.4 (-0.6 to -0.2)	0.001	20%

Data are presented as mean ± SD and mean differences (95% CI); p value when obtained from paired samples t-test.

^aOne of 30 patients did not attend muscle strength tests, leaving 29 pairs for this analysis.

Fig. 2.

A-C Subgroup analysis, based on reconstruction site (proximal femur, n = 11; distal femur, n = 14; proximal tibia, n = 4), of isometric muscle strength (Nm/kg) in (A) knee extension; (B) knee flexion; and (C) hip abduction. Muscle deficits (that is, percent decrease in muscle strength) in nonsurgical and surgical limb compared with controls (above bars) and in surgical limb compared with nonsurgical limb (below bars). Error bars represent mean (95% CI). ^aStatistical significance (p < 0.05), ^bstatistical significance (p < 0.001), when obtained from paired sample t-test.

Pain intensity ratings differed between groups for “worst pain” and “mean pain,” but not for “current pain” (Table 4). “Worst pain” showed a median (interquartile range) of 5 (5) and 0 (2) points (p = 0.001), “mean pain” 3 (3) and 0 (1) points (p = 0.002), and “current pain” 1 (3) and 0 (1) points (p = 0.07) in patients and controls, respectively (Table 4). Eighty-seven percent (26 of 30) of patients marked pain in at least one of 42 body regions on the pain drawing. The patients reported a median (IQR) of 2.5 (4) painful body regions on the pain drawings (Table 4), with knee, thigh, and gluteal area as the most frequent painful body regions (Fig. 3).

Table 4.

Between-group comparison of pain intensity and number of painful body regions (of maximum 42 body regions)

Pain		Patients (n = 30)	Controls (n = 30)	p value
Intensity (NRS)	Current pain	1 (3)	0 (1)	0.07
	Worst pain during the past week	5 (5)	0 (2)	0.001
	Mean pain during the past week	3 (3)	0 (1)	0.002
Distribution	Number of painful body regions	2.5 (4)	0 (3)	0.01

Data are presented as median (interquartile range); NRS = numeric rating scale (0 = no pain to 10 = worst pain imaginable).

Fig. 3.

Painful body regions marked by two or more of 30 patients after limb-sparing surgery with reconstruction of proximal femur (n = 12), distal femur (n = 14), and proximal tibia (n = 4). We modified a template from a previous paper (Adapted with permission from Wolters Kluwer Health Inc: Lacey RJ, Lewis M, Jordan K, et al. Interrater reliability of scoring of pain drawings in a self-report health survey. Spine (Phila Pa 1976). 2005;30:455-458.)

ICF Activity and Participation

Group differences were found for the 6MWT, 30-second CST, and TESS (Table 5). Mean 6MWT was 499 ± 100 meters versus 607 ± 68 meters (mean difference -108 meters; p < 0.001), mean CST was 12 ± 5 repetitions versus 18 ± 5 repetitions (mean difference -7 repetitions; p < 0.001), and median (IQR) TESS scores were 78 (21) points versus 100 (10) points (p < 0.001) in patients and controls, respectively (Table 5).

Table 5.

Between-group comparison of the 6MWT, CST, and TESS

	Patients	Controls	Between groups	p value
6MWT, m	499 ± 100	607 ± 68	-108 (-146 to -70)	< 0.001
CST, n	12 ± 5	18 ± 5	-7 (-9 to -4)	< 0.001
TESS	78 (21)	100 (10)	-14 (-26 to -5)	< 0.001

Data are presented as mean ± SD for 6MWT and CST, median (interquartile range) for TESS, and as between-group mean differences (95% CI); 6MWT = 6-minute walk test; CST = 30-second chair-stand test; TESS = Toronto Extremity Salvage Score (range 0-100). CST measures complete chair-stand movements.

Quality of Life

There were differences between groups in the EORTC QLQ-C30 subscales global health (median [IQR] 79 [16] versus 92 [17]; p < 0.001), physical functioning (median [IQR] 80 [30] versus 100 [0]; p < 0.001), role functioning (median [IQR] 83 [50] versus 100 [0]; p < 0.001), fatigue (median [IQR] 22 [33] versus 11 [14]; p = 0.01), and pain (median [IQR] 25 [33] versus 0 [17]; p < 0.001) (Table 6).

Table 6.

Between-group comparison of the EORTC QLQ-C30 subscales

EORTC QLQ-C30		Patients (n = 30)	Controls (n = 30)	p value
Global health status		79 (16)	92 (17)	< 0.001
Functional scales	Physical functioning	80 (30)	100 (0)	< 0.001
	Role functioning	83 (50)	100 (0)	< 0.001
	Emotional functioning	92 (25)	96 (17)	0.37
	Cognitive functioning	100 (17)	100 (17)	0.63
	Social functioning	100 (33)	100 (0)	0.001
Symptom scales	Fatigue	22 (33)	11 (14)	0.01
	Nausea and vomiting	0 (0)	0 (0)	0.18
	Pain	25 (33)	0 (17)	< 0.001
Item scales	Dyspnea	0 (33)	0 (0)	0.22
	Insomnia	0 (33)	0 (33)	0.31
	Appetite loss	0 (0)	0 (0)	1.0
	Constipation	0 (0)	0 (0)	0.66
	Diarrhea	0 (0)	0 (0)	0.06
	Financial difficulties	0 (33)	0 (0)	0.003

Data are presented as median (interquartile range); the EORTC QLQ-C30 scales range from 0 to 100 points; higher scoring represents better status for Global health status and Functional scales; higher scoring represents worse status for Symptom and Item scales; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire.

Correlation between ICF Body Function, Activity and Participation Outcomes, and QoL

Overall, muscle strength, pain, and TESS correlated to QoL scores (Table 7). Worse global health status correlated to less hip abduction muscle strength in both surgical (Rho 0.43; p = 0.02) and nonsurgical limbs (Rho 0.39; p = 0.04) and higher current pain (Rho -0.60; p < 0.001) (Table 7). Worse physical functioning of the EOTRC QLQ-C30 correlated to less muscle strength in knee extension (Rho 0.40; p = 0.034), knee flexion (Rho 0.40; p = 0.038), and hip abduction (Rho 0.51; p = 0.006) of the surgical limb and hip abduction (Rho 0.38; p = 0.046) of the nonsurgical limb (Table 7). Worse physical functioning also correlated with higher pain scores (current pain, Rho -0.54; p = 0.002; worst pain, Rho -0.51; p = 0.004; mean pain, Rho -0.40; p = 0.03; and painful body regions, Rho -0.53; p = 0.003) (Table 7). Better scorings of the subscale physical functioning correlated with better scorings of the TESS (Rho 0.79; p < 0.001) (Table 7).

Table 7.

Relationship between quality of life and variables of ICF body function and ICF activity and participation (n = 29)

		EORTC QLQ-C30
		Global health	p value	Physical functioning	p value
ICF body function
ROM	Hip flexion (n = 11)a	0.16	0.64	0.03	0.92
	Hip extension (n = 11)	-0.21	0.53	0.15	0.67
	Hip abduction (n = 11)	-0.29	0.40	0.46	0.16
	Knee flexion (n = 18)	-0.31	0.21	-0.28	0.28
	Knee extension (n = 18)	-0.18	0.48	-0.37	0.14
Muscle strength in surgical limb	Knee extension	0.15	0.43	0.40	0.034
	Knee flexion	0.16	0.42	0.40	0.038
	Hip abduction	0.43	0.02	0.51	0.006
Muscle strength in nonsurgical limb	Knee extension	0.31	0.10	0.31	0.11
	Knee flexion	0.39	0.04	0.35	0.07
	Hip abduction	0.39	0.04	0.38	0.046
Pain intensity (NRS)	Current pain	-0.60	< 0.001	-0.54	0.002
	Worst pain	-0.36	0.05	-0.51	0.004
	Mean pain	-0.31	0.09	-0.40	0.03
Pain distribution	Painful body regions	-0.38	0.04	-0.53	0.003
ICF activity and participation
6MWT		0.01	0.98	0.20	0.32
CST		0.19	0.32	0.29	0.14
TESS		0.38	0.046	0.79	< 0.001

Data are presented as correlation coefficients (Rho). The CST measures complete chair-stand movements; EORTC QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire; NRS = numeric rating scale; 6MWT = 6-minute walk test; CST = 30-second chair-stand test; TESS = Toronto Extremity Salvage Score.

Discussion

Limb-sparing bone tumor resection and reconstruction with a tumor prosthesis is an extensive surgical procedure. In the postoperative period, limitations in self-reported physical function and inferior QoL have been reported [54]. Objectively measured physical function has been suggested to serve as an important supplement to provide a more complete image of physical status after surgery [19]. Moreover, evaluating the relationship between different aspects of physical function and QoL can help health professionals involved in rehabilitation to target intervention to improve QoL. In this study, we found that the patients who had undergone limb-sparing surgery 2 to 12 years earlier had bilateral muscle deficits in knee extension and hip abduction and less walking and chair-stand capacity. The patients reported weekly pain most frequently located in thigh, knee, and gluteal area. To give tangible expectations about future physical status, information about frequent pain, reduced muscle strength, and mobility capacity should be provided to patients who face this surgery. In addition, this study showed that muscle strength and pain were related to QoL, thus indicating that these two aspects of physical function are important for the perception of QoL and can be targeted in rehabilitation.

Limitations

This study has several limitations. First, the sample size was small; however, because bone sarcoma is a rare disease, this was expected [60]. Of the 44 eligible patients 32% (14 of 44) were not included. Reasons for not participating varied: Two patients were in late-stage palliative care, and two patients thought the transportation to the study location was too challenging. With the exclusion of these four patients, a possible selection bias toward better outcomes was introduced. The seven patients who declined or did not return calls we assumed were less motivated to attend. The results would therefore be generalizable to survivors of bone sarcoma after limb-sparing surgery with the motivation and overall physical status to attend outpatient clinics and endure physical testing. The attendance rate of 68% gives surgeons an idea of the percentage of patients who regain mobility and the motivation to participate in physical activity at a reasonable level. Second, a heterogenous sample of patients with different sites of surgery, resection length, histology, (neo-) adjuvant treatment, assessment points between 2 and 12 years after surgery, and ages ranging from 19 to 83 years was included. However, we decided to pool data from these patients based on the pre-analyses, which showed no differences in TESS scores between the three sites of surgery and no correlations between the TESS scores and time from surgery or age. The absence of correlation between the TESS scores and time from surgery is further supported by a study that followed children after limb-sparing surgery who showed no improvements in TESS between 2 and 7 years postoperatively [59]. Furthermore, a study assessing TESS and QoL in adults at a mean 4 years after surgery found no correlations between time since surgery and score ratings [33]. We therefore considered pooled analyses for this sample to be legitimate. Third, the study was cross-sectional in design because we only assessed patients at one point. Therefore, we cannot draw any conclusions about the temporal relationship of time of surgery and onset of the impairments and disabilities found in this study. Forth, blinding test personnel to the operated surgical joint was not possible. This may have encouraged patients to perform better so as to demonstrate good surgical results. To minimize observation bias, we used standardized information, instructions, and encouragement, and conducted training sessions before the study started. Because they showed clinically important reductions, we believe the risk of overestimating patients’ results did not affect the interpretation of between-group differences. Despite limitations in study design and a possible skewness toward overestimating outcome scores, we believe these results are important. One reason is to give the most accurate information possible to patients before their consent to limb-sparing surgery. Each patient has different needs for information, and health professionals should be prepared to supplement standard information with personal advice to suit the individual’s situation [25]. Another reason is that the use of objectively measured physical function of a generic nature enables comparison to population norms and between different treatment modalities. The assessments used in this study can be easily implemented in clinical practice. A third reason is to contribute to research. Considering the challenges of conducting prospective studies in a rare disease with 5-year survival rates of about 60%, the cross-sectional design has its place. However, to evaluate the course of function over time, prospective studies are needed.

Differences Between Patients and Controls

ICF Body Function

The surgical limb showed muscle strength deficits in all three muscles groups tested regardless of the target joint of surgery. Reduced muscle strength is expected in resected muscles at short- and long-term [8, 12], and normalized muscle strength is generally expected in intact muscles after completing postoperative rehabilitation. Our results showed that, in accordance with expectations, resected muscles were weaker, but in contrast to expectations, intact muscles were also weaker compared with controls. The deficits in hip abduction and knee extension in both surgical and nonsurgical limbs are clinically important [41, 42, 49], and they are larger than in patients with primary knee replacements [27, 43, 58]. We cannot draw any conclusion about the temporal relationship of time of surgery and onset of muscular weakness in the nonsurgical limb. The findings, however, suggest that it would be advisable to assess muscle strength bilaterally, comparing them with reference groups and where muscle deficits are found, to consider strengthening exercises. Hip and knee flexion ROM in this sample was less versus controls but compared with the mechanical abilities of the tumor prosthesis, we would consider mean hip flexion of 113° ± 18° and mean knee flexion of 113° ± 29° acceptable. Our results are similar to one study in children after limb-sparing surgery in the hip and knee [40].

ICF Activity and Participation

We found clinically important reductions in 6MWT, CST, and TESS, and when comparing our patients (mean age 51 ± 18 years) to population-based norms, mean walking and chair-stand capacity were equal to levels of 80-year-old men and women [6, 10, 42, 44, 52, 55]. Further deterioration of walking and chair-standing capacity may have consequences on self-sufficiency, and it has been found to be related to increased risk of falling, disabilities, and mortality in older people [11, 36]. It is unrealistic to think that walking and chair-standing capacity could be as good as age-related references after this extensive surgery. The walking capacity of children and adolescents was demonstrated to improve during the first 2 postoperative years and to level out between 2 and 7 years, but it was still inferior to age-related peers [5, 59]. This may also be the case for adults, but there are no prospective studies to confirm this. Bear in mind that with a capacity of an 80-year-old, it might be advisable to promote exercise to minimize deterioration of these activities over time.

Quality of Life

The patients showed poorer QoL in functioning subscales, fatigue, and pain but not in the cancer-specific item scores (dyspnea, insomnia, appetite loss, constipation, diarrhea). This may be due to initial treatment being completed before inclusion in this study and side-effects from treatment, such as chemotherapy, were no longer an issue. Comparing our results with population-based references, patients’ scores for global health were at equal levels and physical functioning were at levels of 70- to 79-year-old patients, despite having a mean age of 51 ± 18 years [32]. This demonstrates the importance of clinicians paying attention to physical function and the ability to participate in social settings 2 years or more postoperatively. Among patients who express difficulties, supportive interventions might be of value.

Correlation Between ICF Body Function, Activity and Participation Outcomes, and QoL

We found that poorer QoL correlated to muscle strength deficits, pain, and lower TESS scores but not to ROM, 6MWT, or CST. This indicates that, for the purpose of improving QoL, elements of strengthening exercises and pain management may be useful but need to be evaluated in prospective studies. A lack of correlation between QoL and ROM, 6MWT, and CST was unexpected since previous studies have found correlations in cancer and knee replacement populations [20, 28, 38, 40]. This may be due to the discrepancy between subjective and objective measures of the same construct. For example, the reporting of being able to walk long distances is from the beholder’s perception, while measuring walking distance is standardized to a metric system. Lack of a demonstrated correlation with QoL is not equal to being of no importance, merely the fact of not being important for the construct QoL.

Conclusion

Our study showed that patients with bone sarcoma or giant cell tumor of bone who had undergone bone resection and reconstruction with a tumor prosthesis of the hip or knee 2 to 12 years previously had clinically important muscle weaknesses not only in resected muscles around the reconstructed joint but also in the ipsilateral hip for knee reconstructions and in the contralateral limb. Many patients reported pain, and they showed clinically relevant reductions in walking and chair-stand capacity that were comparable to age groups 20 to 30 years older. Objective outcome measures of physical function supplement self-reported outcomes, and together they can give patients facing this type of surgery a more tangible image of long-term physical status. The results are highly relevant in the information before surgery and informed patient consent. These results also showed that several different aspects of physical function were affected, and it might be advisable to include objective measures in postoperative assessments. The outcome measures used in this study were easily administered, cheap, and can be implemented in clinical practice. Prospective studies measuring subjective and objective physical function pre- and postoperatively in adult samples are needed. Repetitive measures over about 5 years could inform us about the course of physical function, enable comparisons to population norms, and help to better design targeted and timely postoperative interventions.