Impact of Physical Rehabilitation on Bone Biomarkers in Non-Metastatic Breast Cancer Women: A Systematic Review and Meta-Analysis

Rehabilitation might improve bone health in breast cancer (BC) patients, but the effects on bone biomarkers are still debated. Thus, this meta-analysis of randomized controlled trials (RCTs) aims at characterizing the impact of rehabilitation on bone health biomarkers in BC survivors. On 2 May 2022, PubMed, Scopus, Web of Science, Cochrane, and PEDro were systematically searched for RCTs assessing bone biomarker modifications induced by physical exercise in BC survivors. The quality assessment was performed with the Jadad scale and the Cochrane risk-of-bias tool for randomized trials (RoBv.2). Trial registration number: CRD42022329766. Ten studies were included for a total of 873 patients. The meta-analysis showed overall significant mean difference percentage decrease in collagen type 1 cross-linked N-telopeptide (NTX) serum level [ES: −11.65 (−21.13, −2.17), p = 0.02)] and an increase in bone-specific alkaline phosphatase (BSAP) levels [ES: +6.09 (1.56, 10.62). According to the Jadad scale, eight RCTs were considered high-quality studies. Four studies showed a low overall risk of bias, according to RoBv.2. The significant effects of rehabilitation on bone biomarkers suggested a possible implication for a precision medicine approach targeting bone remodeling. Future research might clarify the role of bone biomarkers monitoring in rehabilitation management of cancer treatment induced bone-loss.


Introduction
Breast cancer (BC) is the most common malignancy in women, with an increasing incidence worldwide [1]. In the last years, the mortality rate related to BC significantly decreased due to the advances in screening programs, early diagnosis, and therapeutical interventions [2]. However, in response to the progressive increase in BC survivors, the prevalence of long terms disabling consequences in these women is steadily increasing, along with the growing need for therapeutic intervention addressing physical and psychosocial impairment that characterizes the so-called "survivorship issues" in BC women [3][4][5]. In this scenario, cancer treatment-induced bone loss (CTIBL) is a common consequence of cancer treatments affecting several BC survivors [6][7][8]. Hormonal therapy (HT) is the gold standard adjuvant therapy for postmenopausal women with hormone receptor (HR)positive non-metastatic BC [6][7][8]. However, HT negatively affects bone mineral density (BMD) due to residual serum endogenous estrogen levels deprivation, leading to a significant increase in fragility fracture risk [8][9][10][11]. Concurrently, chemotherapy has been related to an unspecific increase in bone resorption, while corticosteroids drug administration has been widely documented to have detrimental consequences on bone health due to a reduction in both bone formation and osteoblast and osteocyte viability [12,13]. Therefore, several pharmacological approaches have been proposed to counter CTIBL, with growing evidence emphasizing the need for precise risk stratification to better guide clinicians in anti-resorptive drug prescription to preserve bone health and reduce the risk of fragility fractures [14][15][16]. On the other hand, lifestyle medicine plays a pivotal role in the multicomponent management of bone and muscle health status in non-metastatic BC survivors, with several relevant guidelines recommending the implementation of a comprehensive CTIBL management, including a calcium-enriched diet, oral supplementation of 1000-2000 IU of vitamin D3 daily, and physical exercise to counteract a potential osteosarcopenia [17][18][19].
More in-detail, physical exercise might prevent bone loss, increase BMD, and reduce fall risk due to the well-known improvement in physical function, balance control, and muscle strength [20]. In this context, hip and trunk muscles are considered as main targets for physical training aiming at stimulating exercise-induced osteogenic effects [21]. To date, several studies supported the role of rehabilitation and physical exercise in improving bone health and quality of life in post-menopausal osteoporotic women [22,23]. More in-detail, the recent systematic review and meta-analysis performed by Kemmler et al. [22] underlined that different exercise modalities might positively affect BMD at the lumbar spine, femoral neck, or total hip site in postmenopausal women [22]. However, to date, the role of rehabilitation in preventing and managing CTIBL is far from being fully understood, whereas recent research is now focusing on the implementation of a precision medicine approach to rehabilitation interventions in accordance with the recent trend of biomarkerbased treatment of cancer patients [24,25]. Thus, despite the mechanisms underpinning CTIBL being far from understood in detail, targeting specific molecular modifications might be considered a promising therapeutical approach in the precision medicine management of bone health in BC survivors. On the other hand, evidence supporting precise monitoring of biological effects of rehabilitation interventions is still lacking, not only in cancer patients, but also in other fields of medicine. Moreover, to the best of our knowledge, no previous systematic reviews assessed the effects of different exercise modalities on bone biomarkers in BC survivors.
Therefore, the aim of this systematic review and meta-analysis was to assess the impact of physical rehabilitation interventions on bone biomarker modifications in non-metastatic BC patients. This might potentially guide physicians and future research to more precise monitoring of bone health and CTIBL treatment in these women.

Registration
This systematic review of randomized controlled trials (RCTs) was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement [26]. Preliminary research on the international prospective register of systematic reviews (PROSPERO) was performed to evaluate if other similar works were in progress. No similar review was identified, thus the study was submitted to PROSPERO and accepted on 2 May 2022 (available at https://www.crd.york.ac.uk/prospero, accessed on 16 December 2022, registration number CRD42022329766).

Search Strategy
Five databases on medical sciences and physical and rehabilitation medicine were systematically searched on 10 May 2022. Two investigators independently searched PubMed/Medline, Scopus, Cochrane Central Register of Controlled Trials (CENTRAL), Physiotherapy Evidence Database (PEDro), and Web of Science (WOS). Duplicates were excluded independently by two investigators. Further details of the search strategy are reported in Table 1.

Study Screening and Eligibility Assessment
After duplication removal, two investigators independently reviewed the title and abstracts of the retrieved records to choose relevant articles. Discordances between the two authors were solved by collegial discussion. A third reviewer was asked if consensus was not possible. All the reports that met the inclusion and exclusion criteria were screened in full text by the same investigators, and the records that met the eligibility criteria were included in the data extraction. Any disagreements between the two investigators were discussed with a third reviewer to reach consensus.

Data Extraction and Synthesis
All data were assessed and extracted independently from full-text documents into Word by two authors. Any disagreement between the two reviewers was solved by collegial discussion among the Authors. In case of disagreement, a third author was asked. All the data extracted were summarized in tables.

Meta-Analysis
The meta-analysis was performed by Revman 5.4.0 (The Cochrane Collaboration, 2020, Boston, MA, USA). Changes in serum markers were displayed as mean difference percentage (MD%) and standard deviation (SD). The heterogeneity among comparisons was estimated by the Chi-squared and I 2 statistic tests. An I 2 > 75% determined significant heterogeneity across the articles. In the event of considerable heterogeneity, a randomeffects model was adopted to determine the pooled estimates with the effect size (ES) and 95% confidence interval (CI). Missing means and SDs were estimated from medians, ranges, and interquartile ranges (IQRs) using the method introduced by Hozo et al. [28].

Quality Assessment and Risk of Bias
The quality of the studies included was assessed independently by two authors, according to the Jadad scale [29]. Discordances were solved by discussion between the authors or by asking a third reviewer. The items assessed were the following (i) random sequence generation; (ii) appropriate randomization; (iii) blinding of participants or personnel; (iv) blinding of outcome assessors; and (v) withdrawals and dropouts. A Jadad score between 3 to 5 points was considered high quality.
The Cochrane risk-of-bias tool for randomized trials (RoBv.2) [30] was implemented for risk of bias assessment. The following domains were assessed by RoBv.2: (i) randomization process; (ii) deviations from the intended interventions; (iii) missing outcome data; (iv) measurements of the outcome; and (v) selection of the reported results. According to these items, bias was classified as low, high, or having some concerns.

Results
Through our search strategy, 352 records were identified from the five databases. After duplication removal, 249 studies were assessed for eligibility and screened for title and abstract. Therefore, 220 records were excluded, and 29 full-text records were assessed for eligibility. Nineteen records were excluded for inconsistency with the eligibility criteria (two were only abstracts, eight studies did not assess relevant bone biomarkers, five studies were not RCT, two studies were RCT protocols, one study did not assess a homogeneous sample of BC patients, and one study did not assess rehabilitation intervention). The studies assessed in full text and the reasons for exclusions are presented in detail in Supplementary  Table S1. Lastly, 10 studies were included in the present work [31][32][33][34][35][36][37][38][39][40]. Figure 1 shows the PRISMA 2020 flow diagram of the search process in detail.
Through our search strategy, 352 records were identified from th After duplication removal, 249 studies were assessed for eligibility and and abstract. Therefore, 220 records were excluded, and 29 full-text recor for eligibility. Nineteen records were excluded for inconsistency wi criteria (two were only abstracts, eight studies did not assess relevant b five studies were not RCT, two studies were RCT protocols, one study homogeneous sample of BC patients, and one study did not asse intervention). The studies assessed in full text and the reasons for exclusio in detail in Supplementary Table S1. Lastly, 10 studies were included in [31][32][33][34][35][36][37][38][39][40]. Figure 1 shows the PRISMA 2020 flow diagram of the search pr

Study Characteristics
The RCTs included were published between 2010 [36,38] and 201 nationalities of the studies included in this review were as follows: sev were conducted in the USA [33,[35][36][37][38][39][40], one (10%) was conducted in A (10%) was conducted in Brazil [32], and one (10%) was conducted in Sou the characteristics of the included studies are shown in detail in Table 2.

Participants
In the present review, 873 subjects (100% females) were assessed in the included studies. More in-detail, 442 BC patients were included in the intervention groups, while 431 BC patients were included in the control groups. The ages of the subjects included ranged from 45.2 ± 5.9 years [37] to 66.6 ± 9.6 years [32]. The body composition was assessed by BMI, and it ranged from 23.3 ± 4.3 kg/m 2 [34] to 33.5 ± 5.5 kg/m 2 [33]. However, it should be noted that one study [38] reported the number of patients per range of age (≤60, >60 years) and BMI (≤25, >25 kg/m 2 ).
The cancer stages ranged from 0 [34][35][36]39] to IIIB [36], but it should be noted that one study [31] did not characterize the cancer stage, although including only non-metastatic BC patients.
Hormonal therapy was administered to 100% of study participants in three studies [31,32,35]. Among the other studies, hormone therapy administrations in the intervention groups ranged between 42% [36] and 78.3% [34], while in the control group it ranged between 53.7% [39] and 85% [34]. On the other hand, two studies did not characterize endocrine therapy administration [33,38]. Table 2 shows further details on cancer stage and cancer treatments received in each study included.

Control Groups
Control groups included BC patients that underwent usual care, vitamin supplementation, pharmacological treatment, stretching and relaxation exercises, and/or psychosocial support therapy. More in-detail, rehabilitation treatment was compared to usual care in three studies [31,33,35], standard treatment combined with psychosocial support in one study [36], stretching and relaxation techniques in three studies [32,39,40], monthly health newsletter in one study [37], and pharmacological intervention with risedronate, calcium, and vitamin D administration in one study [38].
The groups have been characterized in detail in Table 3.

Primary Outcome-Bone Biomarkers Modifications
In the present review were included RCTs assessing the biological effect of different rehabilitation programs in terms of modification of concentration of the markers described below.

•
Procollagen type I N-terminal propeptide (P1NP) was assessed in two studies [31,37], reporting significant changes in one study [37]. More in-detail, Tabatabai et al. [37] reported a significant decrease (p < 0.05) in both intervention and control groups.

•
Osteocalcin was assessed in five papers [32,33,37,39,40]; out of these, two studies [32,33] reported a significant increase (both p < 0.05) in the intervention group after CET, while Winters-Stone et al., 2011, [39] reported a significative inter-group difference after RET combined with IET (p = 0.01). Lastly, Tabatabai et al. [37] reported a significant decrease in both intervention (CET) and control groups. • Bone-specific alkaline phosphatase (BSAP) was assessed in four papers [33,35,36,38]; among these studies, Dieli-Conwright et al. [33] reported a significant increase in serum concentration in the CET intervention group compared to intervention; concurrently, Waltman et al. [38] reported a reduction in both the intervention group (RET + risedronate, calcium, and Vitamin D) and control group (risedronate, calcium, and Vitamin D). The remaining studies did not report a significant modification of BSAP values (p > 0.05). • Deoxypyridinoline change in serum level was assessed in two studies [39,40], although neither reported a significative intra-or intergroup difference (p > 0.05).

•
Receptor activator of nuclear factor (RANK) was assessed by Dieli-Conwright et al. [33], but the study did not report significant changes (p > 0.05).

•
Whole body BMD was assessed in two studies [32,33,37], without reporting significant changes in intergroup analysis.  [38], with no significant changes after the intervention. Table 4 summarized the primary and secondary outcomes of the present review.

Meta-Analysis
A meta-analysis was performed to underline the effects of different exercise interventions on bone metabolism biomarkers of non-metastatic BC patients, showing an overall significant MD% decrease in NTX serum level [ES: −11.65 (−21.13, −2.17), p = 0.02)] and an increase in BSAP levels [ES: +6.09 (1.56, 10.62), p = 0.008)]. On the other hand, no significant differences were found for urinary NTX, CTX, and osteocalcin markers. Percentage differences between intervention and control group provided by Waltman et al. [38] were used to adapt the data related to the combined intervention (exercise combined with risedronate versus risedronate alone). A random-effects model was adopted since the low number of RCTs included and the high heterogeneity of rehabilitation intervention (for further details see Figure 3).

Meta-Analysis
A meta-analysis was performed to underline the effects of different exercise interventions on bone metabolism biomarkers of non-metastatic BC patients, showing an overall significant MD% decrease in NTX serum level [ES: −11.65 (−21.13, −2.17), p = 0.02)] and an increase in BSAP levels [ES: +6.09 (1.56, 10.62), p = 0.008)]. On the other hand, no significant differences were found for urinary NTX, CTX, and osteocalcin markers. Percentage differences between intervention and control group provided by Waltman et al. [38] were used to adapt the data related to the combined intervention (exercise combined with risedronate versus risedronate alone). A random-effects model was adopted since the low number of RCTs included and the high heterogeneity of rehabilitation intervention (for further details see Figure 3).
The study by Tabatabai et al. [37] was excluded from the meta-analysis because numerical data were not reported.  The study by Tabatabai et al. [37] was excluded from the meta-analysis because numerical data were not reported.

Quality Assessment and Risk of Bias
According to the Jadad scale, eight (80%) RCTs were considered high-quality studies [31,[33][34][35][36][37]39,40]. Lower quality was found in two (30%) studies [32,38] due to missing information about randomization methods or blindness of data assessors. On the other hand, it should be noted that blindness of participants and personnel was not achievable in all the studies included due to the intrinsic nature of the rehabilitative treatment. Table 5 showed in detail the score of each subitem of the Jadad scale for the RCTs included.  [31,34,39,40] (40%) with a low overall risk of bias. Two studies (20%) [33] showed some concerns in the second domain for deviations from intended interventions due to the missing appropriate analysis for effect of assignment to intervention. These concerns lead to an overall medium risk of bias. Lastly, 1 study [33] (10%) showed a high risk of bias for exclusion of five women from analysis after the intervention, resulting in a high overall risk of bias (see Figure 4).

Discussion
In recent years, the long-term management of BC survivors has gained a rising interest in both clinical and research settings, considering the growing prevalence of cancer disabling sequelae affecting these women. Several papers highlight the need for structured and tailored rehabilitation intervention to improve both physical and psychosocial well-being of BC women [41,42]. In this scenario, CTIBL is widespread disabling condition in cancer patients and physical exercise plays a pivotal role in its prevention due to the multifaceted effects on the whole musculoskeletal system, improving both BMD and reducing the risk of falling in patients at high risk of fragility fracture [43][44][45]. However, to date, several questions are still open about the precise biological effects of physical exercise on bone metabolism and health since the complex multilevel interactions characterizing CTIBL in non-metastatic BC survivors. In light of these considerations, this meta-analysis of RCTs assessed the effects of different exercise modalities on currently available bone biomarkers, providing a broad overview about the evidence supporting biomarker implementation in the clinical setting in order to guide physicians in a precise prescription of individualized rehabilitation plans.
Interestingly, our meta-analysis showed significant effects in terms of NTX serum level [ES: −11.65 (−21.13, −2.17), p = 0.02)]. However, it should be noted that the results of individual studies were not significant in the majority of the RCT included. This limitation might be partly related to the small sample of the studies considered and the low effect size, given that the results of the pooled sample showed significant differences in terms

Discussion
In recent years, the long-term management of BC survivors has gained a rising interest in both clinical and research settings, considering the growing prevalence of cancer disabling sequelae affecting these women. Several papers highlight the need for structured and tailored rehabilitation intervention to improve both physical and psychosocial well-being of BC women [41,42]. In this scenario, CTIBL is widespread disabling condition in cancer patients and physical exercise plays a pivotal role in its prevention due to the multifaceted effects on the whole musculoskeletal system, improving both BMD and reducing the risk of falling in patients at high risk of fragility fracture [43][44][45]. However, to date, several questions are still open about the precise biological effects of physical exercise on bone metabolism and health since the complex multilevel interactions characterizing CTIBL in non-metastatic BC survivors. In light of these considerations, this meta-analysis of RCTs assessed the effects of different exercise modalities on currently available bone biomarkers, providing a broad overview about the evidence supporting biomarker implementation in the clinical setting in order to guide physicians in a precise prescription of individualized rehabilitation plans.
Interestingly, our meta-analysis showed significant effects in terms of NTX serum level [ES: −11.65 (−21.13, −2.17), p = 0.02)]. However, it should be noted that the results of individual studies were not significant in the majority of the RCT included. This limitation might be partly related to the small sample of the studies considered and the low effect size, given that the results of the pooled sample showed significant differences in terms of NTX. NTX is one of the most important biomarkers to assess bone resorption [46][47][48]. Its levels in bloodstream reflect the liberation of peptides produced by degradation of osteoid (composed mostly of collagen); in this context, its serum levels might quantify the rate of bone resorption [49], also considering the role that might play in repairing bone and nerves [50,51]. Moreover, the recent systematic review by Migliorini et al. [52] found a significant association between NTX serum level and lower spine and hip BMD, suggesting that a NTX serum level might reflect an increased bone turnover, leading to a reduction in both BMD and T-score. Interestingly, the qualitative synthesis identified one study [38] reporting significant changes in NTX serum levels after RET intervention, suggesting that RET might be the most promising modality in inducing NTX serum level modifications. However, it should be noted that urinary NTX excretion did not show significant changes in the meta-analysis [37].
In recent years, the International Osteoporosis Foundation (IOF) and the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) identified NTX and CTX as the most promising bone biomarkers in the clinical setting of bone pathological conditions, including osteoporosis [53,54]. Despite positive results being reported in NTX serum level modifications, the results of our meta-analysis did not show significant differences in terms of CTX modifications after physical exercise programs. However, few studies [37,38] assessed CTX serum levels in BC women undergoing physical exercise programs. Therefore, these data might be significantly affected by the low number of studies currently available in the literature.
Considering bone deposition biomarkers, BSAP is a bone marker of bone formation that showed a significant increase after exercise therapy interventions in BC patients (BSAP levels [ES: +6.09 (1.56, 10.62) p = 0.008)]. Approximately 50% of BSAP is produced from the skeletal system in subjects with normal liver function. However, no studies assessing BSAP reported the liver function status of study participants. Thus, to date, CET or RET seems to be the exercise modality most supported in improving blood levels of BSAP in BC survivors [33,38].
Similarly, osteocalcin is selectively secreted by osteoblast and is considered a bone marker to assess bone anabolic activity [55]. In particular, γ-carboxylate osteocalcin has great affinity for hydroxyapatite and is commonly stored in bone tissues [56]. Osteocalcin decarboxylation promotes its endocrine activity as bone-derived hormone, with recent studies highlighting its role in glucose metabolism [55][56][57]. Our results highlighted a significant improvement in terms of osteocalcin serum level in two studies [32,33] assessing CET or RET combined with IET. On the other hand, the meta-analysis did not show significant benefits of exercise in terms of osteocalcin.
In the last two decades, increasing interest has been raised in both RANK and RANK-L pathways, two crucial pharmacological targets in the management of osteoporosis. More in-detail, RANK is a transmembrane receptor involved in the signaling pathway regulating osteoclast differentiation and activation. To date, this pathway is the main target of the monoclonal antibody Denosumab, exerting its antiresorptive action by blocking the interaction between RANK and RANK-L, with consequent inhibition of osteoclast activity [16,58]. On the other hand, RANK is not monitored in the current clinical practice, and there is a lack of studies in terms of modifications after physical exercise training in BC survivors [33].
In addition, RANK monitoring might be crucially affected by the pharmacological therapies commonly administered to prevent CTIBL [59][60][61]. Therefore, RANK and RANK-L should not be considered bona fide biomarkers to assess the biological effects of physical exercise in BC patients. Lastly, no evidence supports their integration in a precision medicine approach focusing on bone health management in BC survivors.
Taken together, our findings showed positive results of certain specific bone biomarkers reflecting the effects of physical exercise on bone health in BC survivors. However, conflicting data were reported about BMD modifications induced by physical exercise in these patients. More in-detail, three studies showed positive results in terms of lumbar spine BMD improvement after physical exercise interventions [37][38][39]. These findings might be probably related to the trabecular structure of vertebra that is metabolically more active and might be more sensible to mechanical stimuli promoting bone formation at the lumbar spine level [62,63]. In addition, RET alone or combined with IET might be the most promising therapeutic approach to improve lumbar spine BMD [37][38][39]. Unfortunately, few studies included in the present work assess T-score or Z-score, probably due to the short-term follow-up period and the non-pharmacological intervention that might provide little changes related to the short terms follow-up and instrumental errors, highlighting another gap of knowledge in the current. In this scenario, previous studies suggested that a multimodal approach, including different exercise modalities, might be the most suitable option to improve bone health in patients with osteoporosis [64][65][66]. Moreover, the recent systematic review by Marini et al. [65] suggested RET and IET as the most promising exercise modalities to reduce the risk of fracture.
On the other hand, several controversies are still open about the macroscopical effects of physical exercise on BMD, and previous systematic reviews and meta-analyses reported insufficient evidence to support a superior effect of one specific exercise modality [67,68]. However, it should be noted that the currently available literature focused on standardized exercise programs without focusing on the biological effects of physical exercise in an individualized rehabilitation plan. Moreover, our systematic review did not identify studies considering a precision medicine approach based on bone remodeling biomarkers to tailor physical exercise programs to the patient's characteristics.
Taken together, our findings underlined that bone biomarkers might be significantly affected by physical exercise and could be possibly implemented in monitoring tailored rehabilitation interventions aimed at treating CTIBL in BC survivors. To the best of our knowledge, this is the first systematic review focusing on the effects of different exercise modalities on bone biomarkers in non-metastatic BC survivors. In the era of precision medicine, a biomarker-based approach might have a role in improving the comprehensive rehabilitation management of these women, including not only physical exercise, but also antiresorptive drugs in patients at high risk of fracture to maximize outcomes and reduce the disability and socio-sanitary costs of fragility fractures [69][70][71]. In addition, due to the widely documented effects of physical exercise on oxidative stress and inflammation, a precise multitarget rehabilitation intervention might not only improve bone health, but also have potential interaction with malignant transformation and tumor progression pathways in BC patients [72][73][74][75][76].
Besides these considerations, we are aware that this study is not free from limitations. More in-detail, the low number of studies included, and the small sample size might limit the strengths of our conclusions. On the other hand, our results reflect the papers currently available about this topic in five different databases and put to light a gap of knowledge in the current literature. However, it should be noted that the sample size assessed allows us to obtain significant results in quantitative synthesis. On the other hand, the heterogeneity of the study population, exercise characteristics, and bone biomarkers might represent the main limitations of the present review. To reduce potential bias related to this issue, we provided a detailed qualitative synthesis to characterize the heterogeneity of the studies. Moreover, meta-analysis has been performed in subgroup analysis for bone biomarkers, limiting the potential implications of their heterogeneity. Lastly, only the study by Waltman et al. [38] assessed the effects of physical exercise in a comprehensive rehabilitation approach to CTIBL, including also pharmacological treatments. In this context, it should be noted that antiresorptive drugs should be integrated into the bone health management of BC survivors receiving AIs in accordance with the most recent guidelines [10,17,18,77]. Given the antiresorptive drugs' effects on bone metabolisms, further good quality studies are needed to better characterize the impact of physical exercise on bone biomarkers in BC patients treated with antiresorptive drugs for preventing CTIBL.
However, our findings might be a catalyst for a deeper understanding of biological processes regulating the multilevel interaction between physical exercise, bone remodeling, and CTIBL. Future research should focus on the precise characterization of physical exercise programs, highlighting the biological differences induced by a comprehensive rehabilitation plan.

Conclusions
Physical exercise is one of the main non-pharmacological interventions counteracting CTIBL in non-metastatic BC survivors. However, to date, no previous systematic review assessed the effects of physical exercise on circulating bone biomarkers, and the effects of different exercise modalities on bone biomarkers are still debated.
Taken together, the results of the meta-analysis suggested significant effects of rehabilitation in terms of NTX and BSAP levels modifications, even though the heterogeneity of the study results might limit the strength of our conclusions. However, our data might have potential implications for the prescription of physical exercise targeting bone remodeling in patients with non-metastatic BC. Future research might clarify the role of bone biomarker monitoring in the comprehensive management of CTIBL to optimize the synergistic role of non-pharmacological and pharmacological approaches in promoting bone health in BC survivors.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.