Increase in major osteoporotic fractures after therapy with immune checkpoint inhibitors

Introduction

Immune checkpoint inhibitors (ICIs) are antibodies that block inhibitory pathways associated with immune activation, which have been proven to be effective at fighting various tumours such as melanoma and lung, breast, gastrointestinal and kidney cancers. Since initial Food and Drug Administration (FDA) approval in 2011, the proportion of patients with cancer in the USA treatable with ICI increased from 1.5% to 43.6%.1 While ICIs have proven to be life saving for many patients, they are associated with immune-related adverse events (irAEs) affecting multiple organ systems including arthritis, myositis, colitis, dermatitis and endocrinologic dysfunction.2 3 Activation of T cells can also lead to bone resorption through the increased production of key bone-resorption cytokines such as receptor activator of nuclear factor kappa-B ligand (RANK-L) and tumour necrosis factor alpha (TNF-α).4 RANK-L and TNF-α trigger proliferation, maturation and activation of osteoclasts leading to bone resorption and, over time, osteoporosis leading to fragility fractures.5

Osteoporotic fractures are associated with decreased quality of life and increased risk of institutionalisation and death.6–8 While the impact of fractures specifically in ICI users is unknown, fractures have recently been shown to increase mortality risk in cancer survivors.7 In spite of the significant morbidity and mortality associated with fractures, osteoporosis screening and treatment are rarely offered to patients with cancer, even after suffering fractures.9

Despite the known effects of T-cell activation on bone, the effect of ICIs on bone has not been elucidated. Two small case series and a pharmacovigilance study have described fractures associated with ICI therapy.10 11 The case series describe a total of six patients who suffered osteoporotic fractures in the absence of diagnosed osteoporosis after ICI initiation.11 The pharmacovigilance study examined 822 drug-event pairs of bone and joint injuries and ICI use from the worldwide pharmacovigilance database Food and Drug Administration Adverse Event Reporting System and found a disproportionality signal of increased spinal compression and hip fractures in patients exposed to ICI compared with patients exposed to all other drugs, suggesting a possible causal relationship.10

A recent population-based study from Canada showed for the first time that the incidence rate of major osteoporotic fractures (MOFs) increased over twofold after ICI initiation.12 However, this study included multiple cancer types, including lung cancers which have a high affinity for bone metastases,13 making it difficult to separate out pathologic fractures from osteoporotic fractures using administrative healthcare data. Their sample size also limited their ability to examine potential predictors of osteoporotic fractures after ICI initiation. A larger sample size with a more homogeneous cancer cohort that is at lower risk of bone metastases would help to confirm the association between ICI use and osteoporotic fractures.

In this population-based cohort study using a large US insurance database, we examine the risk of osteoporotic fractures in patients with melanoma treated with ICI, before and after initiation of ICI treatment.

Methods

Cohort identification

We identified all men and women who were 18 years of age or older and had used at least one dose of FDA-approved ICI during the study period. Patients were required to have an International Classification of Diseases (ICD) 9/10 claims diagnosis of melanoma listed as the primary cancer between 60 days prior to and 60 days after the first ICI usage date (index date). Patients had to have continuous enrolment in the database for a minimum of 18 months prior to the index date and a minimum of 3 months after the index date (figure 1).

Request permission

Figure 1

Cohort selection. ICI, immune checkpoint inhibitor.

Results

The study cohort consisted of 3137 individuals (mean age at ICI initiation 68.0 years, SD 13.8), of which 2010 (64.1%) were men. Table 1 summarises baseline characteristics of the cohort at the index date, stratified by incident MOF status. The mean observation time was similar in both groups (304 vs 289 days, p=0.324). For those who suffered an incident MOF, the average age was older (74.7 years vs 67.9 years, p<0.001), they were more likely to be women (p<0.001) and had slightly less total glucocorticoid exposure. There was a higher percentage of deaths within 1 year of ICI initiation in the group with incident MOF than in the group without (33.3% vs 21.6%, p=0.050). The groups were otherwise similar for year of ICI initiation, type of ICI therapy, race, region, comorbidities, bone metastasis, chemotherapy and radiation exposure (p>0.05). In total, only five individuals had missing values of covariates.

In total, 40 individuals suffered an MOF in the year prior to starting ICI therapy, 57 individuals suffered an MOF in the first year after starting ICI therapy and 34 individuals suffered an MOF in the second year after starting ICI therapy. Time to MOF was significantly shorter in the post-ICI period (log-rank test p=0.003). The HR of suffering an incident MOF in the first post-ICI year as compared with the pre-ICI year was 1.82 (95% CI 1.24 to 2.66, p=0.002). The HR of suffering an incident MOF in the second post-ICI year as compared with the pre-ICI year was 1.85 (95% CI 1.12 to 2.90, p=0.007)

The incidence rate of MOF in the year prior to ICI initiation was 12.8 per 1000 patient-years, which was significantly lower than in the first year after ICI initiation, when it was 24.2 per 1000 patient-years (incidence rate ratio [IRR] 1.90, 95% CI 1.25 to 2.91, p=0.002, table 2). The fracture rate remained elevated in the second year after ICI initiation at 23.9 per 1000 patient-years (IRR 1.87, 95% CI 1.16 to 3.02, p=0.007).

In multivariable models, prior fracture, older age, female sex and combination ICI therapy were associated with greater risk of MOF after ICI initiation (table 3). For each decade of increase in age, the hazard of fracture increased by 58% (HR 1.58, 95% CI 1.21 to 2.07). Having a prior MOF in the year prior to ICI therapy increased the hazard of incident MOF after ICI initiation by 5.48 times. Women were at 2.82 times the risk as compared with men. Combination ICI therapy increased fracture risk by 2.46 times as compared with monotherapy.

Discussion

We found that the risk of suffering an MOF increased by over 80% in the first and second years after ICI initiation in patients with melanoma. Those who were women, of older age, with a history of MOF and exposed to combination ICI therapy were at greater risk of suffering an incident MOF after ICI initiation. Glucocorticoid use, considering either average daily dose or total cumulative dose, was not a significant predictor of MOF after ICI in our multivariable model. However, it should be noted that glucocorticoid use was low in our cohort.

Our primary outcome was consistent with the previous before–after population-based study which found a 2.43 (95% CI 1.34 to 4.27) times increased risk of MOF after initiation of ICI therapy.12 In our study, combination ICI therapy was associated with higher risk of fracture, which is consistent with existing literature that combination ICI therapy is associated with greater risk of irAEs as compared with ICI monotherapy.24 However, in the first study, no difference in risk of MOF was detected between the different types of ICI. This may represent inadequate sample size or a reflection of confounding by indication of underlying cancer type and the association between certain cancer types with increased fracture risk, a potential issue that we were able to mitigate by examining only one cancer type.

Screening tools and treatments to prevent cancer treatment-induced bone loss already exist.25–30 If ICI therapy increases risk of osteoporotic fractures, cancer survivors exposed to ICI therapy may need to be screened and treated for osteoporosis at a lower threshold than the general population. Cancer treatment-induced bone loss guidelines, which currently target those exposed to hormone deprivation therapies, may need to be expanded to include those who have been exposed to ICI.25 27–29 With the success of ICIs in advanced stage cancers, patients are living long enough for post-treatment osteoporosis to become one of the main sources of morbidity and determinants of quality of life. Early identification of those at risk may lead to earlier interventions to prevent bone loss/fractures.

Strengths of this study include adequate sample size to examine predictors of MOF in ICI users while restricting our cohort to one tumour type. We chose melanoma because out of the major tumour types for which ICI therapy is indicated, melanoma is the least likely to metastasise to bone.13 Less than 5% of our cohort had bone metastasis and there was no difference in frequency of bone metastasis between those who fractured and those who did not. Further, we used stringent and previously validated definitions of incident fracture and limited our analysis to MOFs, thereby reducing risk of contamination with other fractures or double-counting prevalent fractures as incident fracture.16 The predictors of MOF found in our study, including female sex, older age and a history of MOF, are consistent with known risk factors for osteoporotic fractures and implies validity in our MOF definitions.31

A limitation of our study is that we could not verify if the MOFs ascertained in this study were osteoporotic fractures or pathologic fractures at sites of bone metastases. This risk was partially mitigated by selecting a cohort of patients with cancer who had a low rate of bone metastasis. We excluded pathologic fracture codes, although misclassification of pathologic and osteoporotic fractures is frequent in administrative claims data.32 While we cannot be certain that all incident fractures in this study were non-pathologic, existing literature from multiple myeloma, prostate cancer, breast cancer and other solid tumours shows that both osteoporotic and pathologic fractures are prevented with antiresorptive bone agents such as zoledronic acid and denosumab and these same agents have been effective at preventing cancer treatment-induced bone loss.27 29 33–36 We were able to examine glucocorticoid use only in the outpatient setting. Thus, glucocorticoids used in hospital would not be captured. Further, the results of this study need to be validated in other cancer groups and populations and additional factors such as body mass index, a family history of osteoporosis, smoking status and alcohol use need to be considered.

Conclusion

MOFs increase after ICI initiation and may represent a new irAE of ICI therapy. Further research is needed to understand the direct causal effects of ICI treatment on bone biology, bone mineral density and fracture risk.