We thoroughly searched multiple databases to find studies on the causes of refractory cancer pain in bone metastases. The electronic databases searched included PubMed, Embase, the Cochrane Library, and Web of Science. The search utilized both keywords and MeSH terms such as “refractory cancer pain,” “bone metastases,” “pain predictors,” “risk factors,” “meta-analysis” and “systematic review.” The Boolean connectors AND and OR were applied to the keywords to enhance the effectiveness of the search strategy.

The articles included in the study were not restricted on the year of publication; however, only articles published in English were considered. Additionally, the reference lists of the identified studies were screened during the database search to identify potentially overlooked articles.

To ensure the relevance and quality of the included studies, the following inclusion and exclusion criteria were used:

All records retrieved from the initial database search were exported into the reference management software EndNote. Duplicates were removed, and the remaining articles were further sifted based on their titles and abstracts to determine study eligibility. This screening process was conducted under the supervision of two independent reviewers, with a third reviewer assisting in cases of disagreement.

After the initial screening based on titles and abstracts, the full texts of the potential studies were obtained and evaluated for inclusion. The final review and meta-analysis included only those studies that met all the inclusion criteria.

Data extraction was carried out independently by two reviewers using a predesigned data extraction sheet. The following information was extracted from each study:

Both reviewers cross-checked the accuracy of all extracted data, and any discrepancies were resolved through discussion.

The risk of bias in the included cohort and case-control studies was specifically evaluated by the Newcastle–Ottawa Scale (NOS). The NOS assesses the quality of studies in three main areas:

These benchmarks were applied to each study, with each receiving a score out of 9 points. Studies were categorized as high quality if they scored 7 or higher, moderate quality if they scored between 5 and 6, and low quality if they scored below 5.

The extracted data were analyzed using statistical analysis using R software. A meta-analysis was conducted to combine the effect sizes of the risk factors across the studies. For each risk factor, OR, HR, or their 95% CI were extracted. Meta-regression was then performed using a random-effects model to address heterogeneity among the studies. Inter-study variability was assessed using I² statistic, which measures the proportion of variation between studies not attributable to pure chance. An I² value greater than 50% was deemed significant enough to indicate heterogeneity. In cases of high heterogeneity, a sensitivity analysis was conducted to identify the potential causes. The presence of publication bias was evaluated using a funnel plot and Egger’s test. If asymmetry was observed in the funnel plot, further investigation was conducted to determine whether the results were due to publication bias or other factors.

A comprehensive search across various databases yielded a total of 856 studies. After applying the inclusion and exclusion criteria, eight studies comprising of prospective observational studies, retrospective studies, pooled analyses, and cohorts were included in this meta-analysis. These studies investigated a range of antecedents and risk factors associated with refractory cancer pain in cancer patients with bone metastases. All studies reviewed in this paper were published between 1996 and 2019, with locations including the USA, China, Japan, Taiwan, and Italy. The population sample sizes varied, with the smallest study involving 31 patients (23) and the largest involving 1,445 patients (24). These studies focused on the antecedents of pain and skeletal complications, pain relief interventions, survival, and the effects of individual clinical interventions in patients with bone metastases. Various cancer types were represented, including prostate, lung, breast, and head and neck cancers; however, particular emphasis was placed on managing refractory cancer pain and skeletal events in patients with advanced or metastatic malignancies (as shown in Table 1 and Figure 1).

The included studies primarily involved middle-aged to older patients, ranging from an average of 58 to 73 years. Most articles analyzed focused on male participants (25–27) while research on breast cancer included female participants (24, 28). Lung, breast, and prostate cancers were the main cancers of interest, especially among metastatic patients. Most of these patients had bone metastases affecting both the axial and appendicular regions. Moreover, studies by Parkes et al. (24) and Cramer et al. (29) examined the various skeletal involvements and their effects on survival and pain control. Notably, it was found that 71% of the breast cancer patients identified in the study by Parkes et al. (24) reported pain symptoms at the time of their bone metastasis diagnosis.

As depicted in Table 2, these studies included in the meta-analysis consistently identified several significant predictors of refractory cancer pain in patients with bone metastases. These predictors include the severity of bone lesions, the nature of the metastasis, the initial level of pain experienced by the patient, and the patient’s performance status. For instance, in patients with prostate cancer, bone pain and the extent of bone involvement were identified as key predictors of SREs, as confirmed by Berruti et al. (26) who found that higher pain scores and greater bone involvement correlated with increased incidence of SREs (HR = 1.13, p < 0.0001, HR = 1.16, p < 0.0001).

Meanwhile, Parkes et al. (24) found that patients with lytic bone metastases had 1.79 times greater odds of experiencing pain compared to those with blastic or sclerotic metastases (95% CI: 1.31–2.43, p = 0.001). In addition, multiple bone metastases were critical in assessing pain levels, given that patients with multiple metastases were more likely to experience pain than those with a single metastasis (OR = 1.37, p = 0.03). Similarly, Tsai et al. (23) identified lesion coverage volume factor as an independent predictor of radiographic response (OR = 1.183, p = 0.006), highlighting the importance of tumor load in determining disease outcomes.

Pain management regimens varied across the studies, with different treatment options being trialed. For instance, Habib et al. (28) examined the effects of chemotherapeutic treatment, radiotherapy, and surgical intervention on acute pain and persistent pain in breast cancer patients, finding that patients with severe acute pain were 5.4 times more likely to develop persistent pain (OR = 5.39, p = 0.001). Patients who underwent post-discharge radiation therapy exhibited a 3.4-fold increase in the likelihood of experiencing persistent pain (OR = 3.39, p = 0.023), confirming the impact of intensive treatments on pain endurance. Halabi et al. (27) studied the effects of opioid medications and pain interference on the survival of patients with metastatic castration-refractory prostate cancer. They found that patients with high pain interference scores faced a 43% increased risk of death, p = 0.001, HR = 1.43. Similarly, Akakura et al. (25) reported that the time for analgesic requirement was an independent predictor of shorter survival, highlighting the importance of effective pain management as a prognostic factor in patients with bone metastases.

Mortality was studied comprehensively in all the included studies, particularly in relation to pain and skeletal events. Parkes et al. (24) found that patients with multiple bone metastases experienced worse overall survival (OS) compared to those with a single bone metastasis (median OS = 4.80 years vs. 7.54 years, p < 0.0001). Moreover, patients with both axial and appendicular skeletal metastases had shorter OS than those with limited metastases confined to either the axial or appendicular skeleton (median OS = 4.58 years vs. 6.78 years, p < 0.0001). Berruti et al. (26) also incorporated the effects of bone metastases on survival, and found that pain levels and alkaline phosphatase were independent predictors of skeletal-related events (SREs). Cramer et al. (29) also uncovered that pain and the quality of life were poorer in patients who underwent trimodality treatment (surgery, radiation, and chemotherapy) and had depression and anxiety (OR of trimodality treatment = 3.55, p < 0.05; OR for depression = 3.91, p < 0.01; OR of anxiety = 4.22, p < 0.01).

Included studies used different pain measurement instruments to measure the intensity of pain and the effect of pain on patients. Numeric Rating Scales (NRS) were most frequently applied, with Tsai et al. (23) and Habib et al. (28) concluding that pain scores of 4 or higher were indicative of refractory cancer pain and worse outcomes. Meanwhile, Halabi et al. (27) utilized the Wisconsin Brief Pain Inventory (BPI) to measure pain interference, which was found to have a significant influence on OS. Furthermore, some studies included physician-assessed pain intensity and patients’ own ratings as the indicators of pain and pain-related interference. For instance, Parkes et al. (24) employed pain medication administration and clinician charting to measure pain at the time of bone metastasis diagnosis.

The majority of the research conducted multivariate analysis to account for various factors, including age, sex, cancer type, and previous treatment. For instance, Tsai et al. (23) used multivariable logistic regression to determine independent predictors of clinical and radiographic response and showed that both Karnofsky performance status and lesion coverage volume were independent predictors. Parkes et al. (24) identified that the presence of multiple bone metastases and concurrent involvement of both axial and limb bones serve as risk factors for poor prognosis, correlating with significantly reduced overall survival rates. Furthermore, the research highlighted additional prognostic factors associated with adverse outcomes, including compromised systemic condition and elevated breast cancer grade, among others. Similarly, Cramer et al. (29) accounted for comorbid conditions and treatment modalities in their assessment of pain outcomes, emphasizing the critical role of patient characteristics in pain management.

Our meta-analysis was conducted using a total of eight studies involving 2,774 patients. The individual incidence rates of refractory cancer pain were shown in Figure 2, Shi et al. (30) reported an individual incidence rate of 18.4%, while Akakura et al. (25) reported a rate of 100%. Utilizing a random-effects model, the pooled incidence was estimated to be 70%. The random-effects model produced a pooled incidence of 70%. This indicates that between 42 and 88% of patients with bone metastases continue to experience uncontrolled cancer pain despite standard management. The current meta-analysis demonstrated a high degree of heterogeneity, with I² = 98%, indicating significant differences in the incidence of refractory cancer pain across the studies. The reason for high heterogeneity may be due to insufficient and missing data from grey research literature or unpublished studies. The τ² value of 2.7198 further supports the presence of variability among the studies included in the analysis.

The common-effect model yielded a slightly lower pooled estimate of 61% (95% CI: 59 to 62%). The significant difference in results between the random effects and fixed effects models suggests substantial heterogeneity, indicating that the true incidence of refractory cancer pain may be influenced by study characteristics such as the cancer type, treatment modalities, and patient population.

The funnel plot for incidence (Figure 3) exhibited some degree of asymmetry, potentially pointing to publication bias or other forms of small study effects. The primary analysis of our meta-analysis identified a higher incidence rate in studies with smaller sample sizes, such as the one by Akakura et al. (25), which reported an incidence rate as high as 91% among 48 patients. In contrast, studies with large sample sizes, such as Parkes et al. (24), which included 1,445 patients, reported more moderate incidence rates of 71%.

In evaluating risk factors, the current cross-study meta-analysis identified 12 risk factors (Figure 4) across the included studies, and subgroup analyses were conducted for each factor. The random-effects model for the subgroup analysis yielded a pooled OR of 4.41 (95% CI: 3.60 to 5.41) for all the risk factors combined, indicating that the presence of these risk factors in patients with bone metastases increases the likelihood of developing refractory cancer pain.

Among the identified risk factors, severe acute pain showed the strongest association with refractory cancer pain, with an OR of 219.2 (95% CI: 0.26 to 188127.63), as reported by Habib et al. (28). However, the broad confidence interval indicates a high degree of variability in this estimate, likely due to small sample sizes and variability in measurement methods. Another notable predictor was breakthrough pain with an OR of 16.44 (95% CI: 0.60 to 448.07), according to Shi et al. (30), but this figure also demonstrated significant variation.

According to Parkes et al. (24), multiple bone metastases emerged as the most consistently reported risk factor across various studies, with an OR of 3.94 (95% CI: 2.64 to 5.87). The authors concluded that patients with extensive bone metastases are approximately 3.72 times more likely to experience refractory bone pain compared to those with limited bone lesions. Additionally, lytic bone metastases had a similarly strong association, with an OR of 5.99 (95% CI: 3.17 to 11.30), indicating that this type of bone lesion significantly increases the risk of refractory cancer pain.

In terms of comorbidities, depression and anxiety were both identified as significant predictors of refractory cancer pain, with OR of 3.91 (95% CI: 1.22 to 2048.64) and 4.22 (95% CI: 1.22 to 2048.64), respectively, according to Cramer et al. (29). These findings suggest that pain evaluation should extend beyond physiological factors, as these may exacerbate pain perception and diminish the effectiveness of cancer treatment. The funnel plot in Figure 5 illustrates the study’s publication bias.

Among the studies included in this analysis, the authors identified the extent of bone metastases as one of the most significant determinants of pain. They found that patients with multiple bone metastases had a higher risk of experiencing refractory cancer pain compared to those with a single metastasis. For instance, Parkes et al. (24) showed that patients with multiple bone metastases were 1.37 times more likely to report pain than patients with single metastasis (95% CI; 1.02 to 1.84, p = 0.03). In a similar vein, Tsai et al. (23) identified lesion volume and Karnowski performance status as the most effective predictors of clinical and imaging response to treatment, thereby highlighting the significance of metastasis burden in relation to pain.

The results from these studies align with the existing knowledge that the extent and location of bone metastases are key determinants of pain and SREs. This is in tandem with the observation that multiple metastases are more destructive to bones as compared to single metastases, with the degree of destruction correlating with pain and other complications. Furthermore, metastases in both axial and appendicular skeleton, as described by Parkes et al. (24) have been shown to predict poor survival rates, highlighting the importance of comprehensive skeletal evaluation in clinical practice.

Another significant finding established in this meta-analysis is the prognostic value of pain in patients with bone metastases. Several studies have described the impact of pain intensity on survival, revealing that high pain intensity is associated with decreased survival. For instance, Halabi et al. (27) observed that patients with high pain interference had a significantly higher HR for mortality (HR = 1.43; p = 0.001), while Akakura et al. (25) found that time to requirement of analgesic was related with poorer survival (p < 0.01). This evidence suggests that pain not only reflects the burden of disease but may also serve as an indicator of disease progression that necessitates more intensive treatment in patients experiencing severe pain.

These findings are consistent with other research that has demonstrated a strong correlation between pain and survival in patients with metastatic cancer (11). Cancer-related bone pain is recognized as resulting from both local bone pathology and the overall tumor burden, as well as inflammation and immune response (31). Tumor-related pain encompasses both local and systemic factors, presenting challenges in the management of cancer pain and highlighting the need to improve patient outcomes.

The meta-analysis indicated that treatment modalities significantly influence pain-related outcomes. Based on randomized samples, studies focusing on patients undergoing aggressive treatments, such as chemotherapy and radiotherapy, demonstrated an increased likelihood of experiencing persistent or severe pain following these interventions. Habib et al. (28) observed a significant association between acute severe pain experienced during breast cancer surgery and the development of persistent pain (OR = 5.39, p = 0.001). Additionally, patients undergoing radiation therapy were found to be at an increased risk of long-term pain (OR = 3.39, p = 0.023). These findings underscore the importance for clinicians to incorporate pain management considerations into treatment planning, particularly for patients undergoing aggressive or multiple therapeutic interventions. Opioids were prescribed as the primary analgesic for patients experiencing severe pain, as documented by Halabi et al. (27) and Akakura et al. (25). Nonetheless, the reliance on opioids underscores the necessity for complementary or adjunctive therapies in the management of patients with challenging pain conditions. Pain management in patients with bone metastases often involves multimodal treatments, including the use of bisphosphonates and other bone-modifying agents. Notably, studies by Parkes et al. (24) and Tsai et al. (23) have demonstrated the efficacy of bisphosphonates and denosumab in preventing SREs and improving pain management.

In the included studies, several variables were found to be significant risk indicators for patients with bone metastases experiencing persistent or severe pain. These included acute pain, the degree of bone involvement, performance status, comorbidities, and duration of pain. Notably, Cramer et al. (29) found that depression and anxiety were associated with more severe pain (OR of depression 3.91, p < 0.01; OR of anxiety = 4.22, p < 0.01). This finding underscores the importance of early intervention for psychological disorders that may heighten patients’ perception of pain prior to the initiation of cancer treatment.

In addition, we found that the type of bone metastases (lytic or blastic) could help predict the severity of pain experienced by patients. Parkes et al. (24) observed that patients with lytic metastases were 1.79 times more likely to report pain than patients with blastic metastases (95% CI 1.34–2.39, p = 0.001). Lytic lesions are known to be associated with more profound bone resorption, resulting in increased pain and decreased structural integrity of the skeletal system, hence the higher pain scores of the patients.

The findings of this meta-analysis yield several important clinical implications. First, identifying potential candidates for pain-related predictors in patients with bone metastases can assist clinicians in the initial evaluation and prevention of pain. Factors such as skeletal metastases, lytic lesions, and poor performance status should be utilized to identify patients at higher risk of developing pain; thus, patients exhibiting these characteristics should be considered for more intensive pain management strategies. Furthermore, the strong correlation between pain and survival underscores the necessity of effective pain management in patients with advanced cancer, as inadequate pain control may indicate disease progression and poor prognosis.

Bone-modifying agents like bisphosphonates and denosumab are used for pain relief and managing SREs (32). These agents are effective not only in preventing pathological fractures and spinal cord compression but also in controlling pain by stabilizing bone structure (33). Therefore, it is essential to incorporate these agents into standard management strategies for patients with bone metastases to enhance both the duration and quality of life.

Lastly, the appropriate management of psychological disorders, such as depression and anxiety, is crucial for individuals experiencing cancer pain. According to Cramer et al. (29), mental health issues significantly contribute to psychological distress, which can influence pain intensity. Addressing these disorders through various interventions such as counseling, medications, or cognitive behavioral therapy can enhance not only pain management but also the overall well-being of the patient.

Despite the valuable insights provided by this meta-analysis regarding the predictors of refractory cancer pain in patients with bone metastases, several limitations should be acknowledged. The included studies were different in terms of patient populations, cancer types, and treatment approaches, raising concerns about the generalizability of the results. Furthermore, most of the included studies were cross-sectional and predominantly descriptive in design, which may limit the ability to draw direct comparisons between risk factors and pain outcomes. Therefore, there is a need to conduct additional randomized controlled trials (RCTs) to establish pain management interventions for patients with bone metastases. Specifically, further research comparing the effectiveness of non-opioid treatments, including bisphosphonates, denosumab, and radiopharmaceuticals, on pain and survival is warranted. In addition, it is essential to thoroughly investigate the role of psychological and social factors in pain management, as these factors are crucial to understanding pain and patients’ overall experiences.