Christoper Wallis: Hello, and thank you for joining us for this Uro Today journal club discussion. Today we're talking about a recent publication entitled Association of Bone Mineral Density Testing With Risk of Major Osteoporotic Fractures Among Older Men Receiving Androgen Deprivation Therapy to Treat Localized or Regional Prostate Cancer. I'm Chris Wallis, an assistant professor in the division of urology at the University of Toronto. With me today is Zach Klaassen, an assistant professor in the division of urology at the Medical College of Georgia.

This is the citation for this recent publication in JAMA Network Open. Prostate cancer is common, representing therefore both a urologic and public health issue as a result of nearly quarter million new cases annually in the US. Androgen deprivation therapy is a mainstay of treatment for many patients with prostate cancer, including both locally advanced disease and metastatic disease. As a result of both the patient population and use of ADT, osteoporosis is common as it is prevalent in this age group and associated with the mechanism of androgen deprivation.

You can see here the relationship between ADT and bone health, through both effects on bone turnover and body mass changes, which lead to an increased fracture risk among patients receiving ADT. And as a result, guidelines recommend screening for osteoporosis. However, historically rates of this have been quite low.

The aim of the present study was to assess contemporary practice patterns regarding the use of DXA scans in men on ADT and fracture risk using both the SEER and Texas Cancer Registry and Medicare-linked data sets.

Now, the authors performed a retrospective cohort study using these registries and identified adult men aged 66 years and older who had histologically confirmed prostate cancer diagnosed between January, 2005 and December, 2015. They restricted their cohort to those who had localized or regional disease according to SEER, and included men who underwent orchiectomy or medical androgen deprivation therapy within 30 days prior to, or one year after their diagnosis. They required that patients not have HMO coverage in the two-year window around their diagnosis and to survive at least one year after diagnosis.

They specifically excluded men with distant or metastatic disease to avoid misclassifying [inaudible 00:02:24] events due to pathologic fractures, when we sought to examine here osteoporotic related fractures. They further excluded men who had fractures at the same anatomic location within one year prior and one year after initiating ADT. It's likely that these represent preexisting events rather than those related to the treatment they're receiving.

They examined a number of outcomes, including the frequency of DXA screening. They defined these based on HCPCS codes and looked at the window from 12 months prior to their first ADT claim to six months after this period. This was done because Medicare reimburses for DXA scans every 24 months in the preventative care setting and every 12 months in high-risk patients. So patients who have recently undergone a DXA scan before initiating ADT may not be eligible for quite some time afterwards.

The authors further examined rates of osteoporotic fractures, including both any fracture and major osteoporotic events. And these were defined as fractures of a hip, other femur, lower arm, upper arm, and spine. And they, again, just to reiterate, excluded those with fractures in the 12 months prior to the first ADT claim. The authors additionally assessed the use of bone-modifying agents, as well as overall survival.

They examined a number of covariates, including important demographic characteristics, such as age, race and ethnicity, marital status, area of residence, state buy-in status, the state of the registry, census level education, and poverty. Additionally, they examined tumor-related variables, including cancer stage, cancer grade, which was dichotomized into low versus high, year of ADT initiation, and ADT type. Finally, they looked at comorbidity by examining prior diagnoses of osteoporosis and prior fractures, as well as a general measure of comorbidity using the Charlson comorbidity index.

They examined DXA screening rates and calculated a 95% confidence interval using a binomial distribution. Multi-variable logistic regression models were used to identify variables that were associated with DXA screening, using a backward reduction selection method, and the final included model had year of ADT, cancer registry, age, marital status, race and ethnicity, cancer stage, cancer grade, Charlson comorbidity score, ADT type, state buy-in status, and prior osteoporosis as predictors.

The time to first fracture was defined from index being the first ADT claim and the authors separately evaluated rates of any major fracture within each fracture site. They used Kaplan-Meier method to visualize this data and propensity score adjusted Cox proportional-hazard models to calculate hazard ratios. Sensitivity analyses were performed restricting to patients who had fractures at least 12 months after their first ADT related claim.

They examined overall survival estimated from the time of first ADT claim to either censoring or death and stratified this according to a fracture. Use of bone-modifying agents was also examined. Among patients who had DXA claims, they looked for the window between the initial DXA screening event and six months after. Whereas for those who did not have DXA claims, the initial window opened with the first ADT claim and closed six months thereafter. I'm going to hand it over to Zach to walk us through the results of this analysis.

Zachary Klaassen: Thanks very much, Chris. So this is essentially the CONSORT diagram for the study and it's quite detailed. And I'll sort of jump down to about halfway through where we see localized original disease, 274,345 patients. And after including patients with first orchiectomy or ADT within 30 days before diagnosis or one year after, this got the patient number down to 73,515, including patients alive at least one year post-ADT, 70,750. And then including patients with Medicare part A B coverage, plus no HMO within 12 months before or after first ADT, 55,346. And then including patients that did not have the same fractures before and after initial ADT, 54,953. And then finally, no fracture in 12 months prior to initial ADT, 53,472 patients included in this analysis.

This is the table one looking at demographic and tumor characteristics for patients who received DXA screening. I've broken this table into two slides, given the size of this table. As we can see at the top here, overall 7.9% of patients in this study had a DXA screen. And if we look at the year of initial ADT, in 2005, it was 6.8% of patients in the study and slight increase up to 8.4% in 2015.

This is the SEER state registry breakdown. We can see here slightly more patients in Hawaii and in New Mexico. And if we look at the age range, we see that older patients had more DEXA scans, upwards of 9.7% for those over 80 years of age. And when we look at marital status, again we see slightly more DXA scans among married patients and slightly less DXA scans among non-Hispanic black patients at 5.2%.

Not surprisingly, with regional stage disease, 11% of patients with DXA scans compared to 7% with localized disease. And in terms of high grade lesions, again, slightly higher at 8.4% compared to 6.3% for low grade cancer. The second half of this table looks at the Charlson comorbidity index. More comorbid patients had more DXA scans, 9.7% for Charlson comorbidity greater than equal to two compared to 7.3% for those with a Charlson comorbidity index of zero.

They included several different types of ADT in this study. And we can see 9.7% of patients receiving Abarelix or Degarelix had a DXA scan, whereas 7.1% of patients receiving leuprolide had a DXA scan. With regards to area of residence, we see slightly less DXA scans among small urban and rural populations compared to big metropolitan and metropolitan residents. And we look at education and poverty level by quartile, one for both of these being the highest level of education, we see more DXA scans in these patients compared to patients with lower education and more poverty.

At the bottom of this table we see that patients that had osteoporosis before their first ADT or DXA scan, they had a DXA scan rate of 22.9%, which you would expect compared to those that did not have osteoporosis. And those that had fracture before first ADT had DXA scans at 13.5%, which was higher than patients that did not have a history of a fracture.

This is the multivariable logistic regression model for association between participant characteristics and DXA screening. In the box I've highlighted on the right is sort of the clinically important features of this model, which I'll go through. We can see here that with the reference of age 66 to 70 years, we see that with increasing age, there is more utilization of DXA screening, with greater than 80 years of age odds ratio of 1.25, which is statistically significant. Again with marital status, married being the reference. We see less DXA scanning with single patients, odds ratio of 0.89. And with non-Hispanic white being the reference, we see a 20% decrease odds of receiving a DXA scan for non-Hispanic black males.

For stage of disease, again, not surprisingly, regional disease, odds ratio 1.44 for receiving DXA scan. High grade cancer odds ratio of 1.25, compared to low grade, and Charlson comorbidity index, again, the sicker patients receiving more DXA scans with Charlson comorbidity index greater than or equal to two with an odds ratio of 1.25 with a reference being Charlson comorbidity index of zero.

The second part of this table, also some interesting points to be made here, again highlighted by this box. Looking at full or partial state buy-in status, less likely to receive a DXA scan with an odds ratio of 0.78 compared to the reference of no state buy-in status. Again, if we look at the education quartiles, with the education quartile one, which is the highest educated, we see significantly decreased odds of receiving a DXA scan for the remaining quartiles, as you can see listed here. Osteoporosis before ADT, significant. And large odds ratio of 16.02 for those with osteoporosis compared to those without.

This is the proportion of patients with fractures after initial ADT according to DXA screening. There was 17.5% of patients in the overall population that had a fracture. And you can see here that those that had a DXA scan had a higher noted incidence of any fracture at 20.2% compared to no DXA scan at 17.3%, which was statistically significant. And if we look at the major fractures, which is defined by spine, upper arm, lower arm, hip, or other femur, again, the patients that had a DXA scan significantly higher notation of DXA scan at 14.9% and 13% for fractures in patients with no DXA scan.

This is the multivariable Cox proportional hazards model for time to first fracture at any site after initial ADT with adjustment for propensity scores. And it's highlighted here at the top of this panel of the table, after adjusting for all of these characteristics, which Chris also mentioned in the methods, there was no difference in time to first fracture for those patients that received a DXA scan with a ratio of 0.96 and a 95% confidence interval of 0.89 to 1.04.

This model looks at the propensity score adjusted multivariable Cox proportional model for time to first major fracture after initial ADT among patients with no previous fracture within 12 months before treatment initiation. And again, after adjusting for these variables, there was no statistical significance for time to first fracture for those receiving a DXA scan with a hazard ratio of 0.91 and a 95% confidence interval of 0.83 to 1.00.

These Kaplan-Meier figures look at overall survival among participants with and without fractures. On the left is any fracture, on the right is major fracture. And we can see very similar lines here with the splitting of the curves for both of these about four to five years after diagnosis or years from initial ADT. And not surprisingly, those patients that did not have a fracture or a major fracture had an improved overall survival, which is statistically significant in both of these models.

This table looks at the association of DXA screening with receipt of bone-modifying agent. I have two boxes here. The small box highlights that only 3.1% of patients in this study received a bone-modifying agent, which is quite low. Not surprisingly, those patients that received DXA scanning were more likely to have bone-modifying agents, 18.8%. And we can see here that this was also consistent when we break down the type of bone-modifying agent, either bisphosphonate or Denosumab.

So several discussion points from this study. This is a cohort study that found a low nationwide rate of 7.9% of DXA screening among older men with localized or regional prostate cancer who started ADT. However, there was small increases from 6.8% utilization of DXA scan in 2005 to 8.4% in 2015. Factors that were associated with receipt of a DXA screening scan included older age, history of osteoporosis or fractures, more advanced or high-risk prostate cancer, and greater number of comorbidities. Demographic and sociodemographic disparities associated with not receiving DXA screening included non-Hispanic black or single males, state buy-in status, which is a surrogate for poverty, living in a census area with lower education levels and higher poverty levels.

Use of the DXA screening to identify patients with indication for bone protective agents is important for bone health management. As I mentioned, only 3.1% of patients received bone-modifying agents in this study. So in conclusion, in this cohort study, the use of DXA screening among older men with localized or regional prostate cancer who initiated ADT remained low and was associated with racial, socioeconomic, and geographic disparities.

Given the deleterious impact of fractures for morbidity and mortality, implementation strategies are needed to increase the uptake of current guidelines for bone health among men with prostate cancer. And finally, early intervention with bone-modifying agents could potentially reduce the burden of illness associated with fractures among older men who are prostate cancer survivors. We thank you very much for your attention, and we hope you enjoy this Uro Today journal club discussion.