Bone fracture as a novel immune‐related adverse event with immune checkpoint inhibitors: Case series and large‐scale pharmacovigilance analysis

Abstract Although immune checkpoint inhibitors (ICIs) are associated with different immune‐related adverse events (irAEs), the potential effect on the skeleton is poorly defined albeit biologically plausible and assessable through pharmacovigilance. We described a case series of patients experiencing skeletal fractures while on ICIs at the National Cancer Institute of Milan. To better characterize the clinical features of skeletal irAEs reported with ICIs, we queried the FDA Adverse Event Reporting System (FAERS) and performed disproportionality analysis by means of reporting odds ratios (RORs), deemed significant by a lower limit of the 95% confidence interval (LL95% CI) > 1. Bone AEs emerging as significant were scrutinized in terms of demographic and clinical data, including concomitant irAEs or drugs affecting bone resorption or causing bone damage. Four patients with skeletal events while on ICIs were included in our case series, of which three exhibited vertebral fractures. In FAERS, 650 patients with bone and joint injuries and treated with ICIs were retrieved, accounting for 822 drug‐event pairs. Statistically significant ROR was found for eight, two and one bone AEs respectively with PD‐1, PD‐L1 and CTLA‐4 inhibitors, being pathological fracture (N = 46; ROR = 3.17; LL95%CI = 2.37), spinal compression fracture (42; 2.51; 1.91), and femoral neck fracture (26; 2.38; 1.62) the most common. Concomitant irAEs or drugs affecting bone metabolism were poorly reported. The increased reporting of serious vertebral fractures in patients without concomitant irAEs and no apparent preexisting risk factors could suggest a possible cause‐effect relationship and calls for close clinical monitoring and implementation of dedicated guidelines.


| INTRODUCTION
The advent of immune checkpoint inhibitors (ICIs) has markedly improved patient survival in different subtypes of metastatic cancer, by enhancing cytotoxic T-cells activity through blocking either cytotoxic Tlymphocyte antigen 4 (CTLA-4) or programmed cell death 1 (PD-1) or its ligand (PD-L1). 1 However, ICIs are associated with a variety of immune-related adverse events (irAEs), virtually affecting all host tissues, most of which have been described through pharmacovigilance analyses. [2][3][4][5][6] The effect on the skeleton is poorly studied and, to the best of our knowledge, only a small case series exists, including three patients with new-onset osteoporosis leading to fracture. 7 This report stems from our experience at the National Cancer Institute Research Center, in Milan, where different cases of suspected ICI-related bone fractures occurred in patients affected by head and neck cancer. This prompted us to investigate the potential biological rationale subtending our findings. Emerging evidence suggests that systemic activation of T cells in vivo leads to an osteoprotegerin ligand-mediated increase in osteoclastogenesis and bone loss ( Figure 1). In fact, ICIs can enhance bone resorption by activating T cells, 8 which in turn causes bone loss with bone fragility, increasing the risk of fractures. 9,10 In the recent past, the Food and Drug Administration Adverse Event Reporting System (FAERS) has attracted considerable interest among clinicians for accurate and timely characterization of drugrelated risks occurring in real-world cancer patients with comorbidities and polypharmacotherapy. These postmarketing studies are particularly suited to early detect rare, unexpected and delayed adverse events (AEs), which cannot be fully appreciated in pivotal trials (where only irAEs occurring in at least 5% of patients were reported), and are recommended for real-time safety assessment of recently marketed drugs receiving accelerated regulatory approval. 11 On these grounds, we aimed to describe spectrum and clinical features of ICI-related skeletal lesions by retrospectively analyzing two real-

| Case and exposure definition in pharmacovigilance analysis
As of March 31, 2020, FAERS collected more than 20 million reports and covered virtually worldwide population (relevant catchment area includes also serious reports from EU and other non-US countries).
We queried the FAERS database (public dashboard) to identify all reports recorded between the first quarter (Q1) of 2004 and Q1 of 2020. We searched all the 112 preferred terms (PTs) listed in "Bone and joint injuries" High Level Group Term (HLGT), and PTs concerning osteonecrosis (namely "osteonecrosis," "osteonecrosis of jaw" and "osteonecrosis of external auditory canal"), classified according to the Medical Dictionary for Regulatory Activities. Furthermore, the event "fall" was searched as negative control, in order to verify whether skeletal toxicity is indirectly related to trauma.

| Disproportionality analysis
As a measure of disproportionality, we calculated the reporting odds ratio (ROR) with relevant 95% confidence interval (CI); statistical significance was defined by a lower limit of the 95% CI of the ROR exceeding 1, with at least 5 cases reported, to reduce the likelihood of false positives. 5 Specifically, a case-noncase approach was applied: cases were defined by "bone and joint injuries" reports recorded for ICIs, while noncases were represented by AE reports recorded for all other drugs in FAERS. The ROR is the odds of exposure to ICIs among the cases divided by the odds of exposure to ICIs among the noncases. If the proportion of the AE of interest is greater in patients exposed to ICIs (cases) than in patients exposed to all other drugs reported in FAERS (noncases), a disproportionality signal emerges. 12,13 Cases counted as many-fold as the number of "bone and joint injuries" events identified by relevant PTs recorded in a given report.

| Clinical characterization of disproportionality signals
Skeletal AEs emerging from disproportionality analysis were further scrutinized to remove potential duplicates (ie, records overlapping in at least three out of four key fields: event date, age, sex, and reporter's country. Remaining cases were described in terms of clinical features, including potential existence of confounders: demographic information (age, gender, reporter country), seriousness (ie, those resulting in death, hospitalization-initial or prolonged-life-threatening events or leading to disability or congenital anomalies), fatality rate (ie, proportion of death reports), therapeutic regimen and indication, concomitant bone metastases, concomitant endocrine irAEs (proxy for occurrence of secondary osteoporosis caused by calcium metabolism disorders, hypogonadism, endogen excess of glucocorticoids or requirement for steroid therapy), proportion of falls and myositis, and concomitant neurological AEs.
Additionally, concomitant drugs were analyzed by searching for proton pump inhibitors (PPIs), suggested to increase the risk of skeleton fracture, 14 agents acting on bone resorption (ie, bisphosphonates, denosumab, teriparatide, as a proxy of preexisting osteoporosis) or causing bone damage (ie, corticosteroids, antiepileptics, antihormonal agents) based on the list proposed by Nguyen et al. 15 Finally, latency of the skeletal events was calculated as the difference between the start of therapy and the date the event occurred (median days with interquartile range-IQR). To avoid the potential confounding factors of concomitant nonskeleton irAEs, the onset was calculated only for cases in which events of interest were reported alone (ie, without concomitant irAEs). The flowchart of methodological steps followed for analysis of FAERS is showed in       The key message of our study is that ICIs may act as precipitating factors for skeletal events. As part of dedicated close monitoring for risk stratification and early detection of skeletal lesions in patients starting treatment with ICIs, laboratory (ie, calcium/phosphorus metabolism) and imaging studies should be performed, also considering the nonnegligible impact of a fracture (ie, immobilization, high-risk of thromboembolic events, increased operative risk) on the quality of life in advanced cancer patients. Furthermore, preexisting osteoporosis/osteopenia, genetic or environmental factors, and concomitant therapies should be carefully considered, including the assessment of body mass index (BMI), due to the potential association between sarcopenia and occurrence of irAEs. 25 Notably, two out of four showed a BMI lower than 18.50. In this context, the implementation of dedicated guidelines for the identification, risk stratification and management of bone lesions in patients receiving ICIs should be pursued.
We acknowledge the limitations of FAERS data, in particular the inability to firmly infer a causal relationship between drug exposure and occurrence of AE. 12 The ROR does not inform the real risk in clinical practice, mainly because of the lack of a denominator and underreporting, but only indicates an increased risk of AE reporting and not a risk of AE occurrence. Therefore, incidence rates and risk ranking cannot be derived from spontaneous reports. Furthermore, the lack of exposure data and clinical elements such as the reporting of preexisting osteopenia/osteoporosis, laboratory and radiological findings makes it difficult to fully evaluate all residual confounders involved in skeletal AEs. Notwithstanding these limitations, pharmacovigilance assessment represents an invaluable opportunity to monitor drug safety and identify novel rare signals, particularly in a setting where ethical and feasibility issues preclude actual conduction of randomized controlled trials.

| CONCLUSIONS
Our large-scale study found increased reporting of serious spinal compression fracture in patients with no apparent preexisting risk factors for skeletal injuries, thus suggesting a possible cause-effect relationship and calling for awareness by oncologists and the implementation of dedicated guidelines. Further investigations are needed to fully characterize this novel irAE, defining patient-and drug-related specific risk factors and optimal management strategies.

ACKNOWLEDGMENTS
This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

DATA AVAILABILITY STATEMENT
The data supporting the findings of this study were derived from the following resource available in the public domain: https://www.fda.gov/ drugs/questions-and-answers-fdas-adverse-event-reporting-systemfaers/fda-adverse-event-reporting-system-faers-public-dashboard. Further details and other data that support the findings of this study are available from the corresponding author upon request.

ETHICS STATEMENT
The study was approved by the Institutional Ethical Committee (INT 216/20, date of approval September 28, 2020). All patients provided written informed consent.