Exploring the Association Between Prostate Cancer Diagnosis per se and use of GnRH Agonists and Diabetes Control in Men with type 2 Diabetes Mellitus: A Nationwide, Population-Based Cohort Study

Background: Gonadotropin Releasing Hormones agonists (GnRH), which are rst line treatment for metastatic prostate cancer (PCa), increase risk of type 2 diabetes mellitus (T2DM). This study aims to quantify the association of use of GnRH with diabetes control in PCa men with T2DM. Methods: Nationwide population-based cohort study in the Swedish National Diabetes Register and Prostate Cancer data Base Sweden 4.1, on the association between GnRH and diabetes control in T2DM men with PCa by comparing T2DM men with PCa vs. without PCa, as well as comparing T2DM men with PCa on or not on GnRH. The primary exposure was use of GnRH. Worsening diabetes control was the primary outcome, dened as: 1) increase of HbA1c to 58 mmol/mol or higher; 2) HbA1c increase by 10 mmol/mol or more; 3) Start of antidiabetic drugs or switch to insulin; 4) combine all denitions above. Cox proportional hazards regression was used to analyze the association. Results: There were 5,714 T2DM men with PCa of whom 692 were on GnRH and 28,445 PCa-free men with T2DM with similar baseline characteristics. Diabetes control was worse in men with GnRH vs. PCa-free men (HR: 1.24, 95% CI: 1.13-1.34) as well as compared with PCa men without GnRH (HR:1.58, 95% CI: 1.39-1.80). Conclusion: Use of GnRH in T2DM men with PCa was associated with worse glycemic control. The ndings highlight the need to closely monitor diabetes control in men with T2DM and PCa starting GnRH agonists and to limit the duration of their use when possible.


Data source
Prostate Cancer data Base Sweden (PCBaSe) 4.1 is a database based on the National Prostate Cancer Register (NPCR) of Sweden, which contains information on 98% of men in Sweden diagnosed with PCa between 1998 and 2016 compared with The Cancer Registry to which reporting is mandated [10]. In PCBaSe 4.1, men in NPCR were linked to other nationwide databases, including National Patient Register, Longitudinal integrated database for health insurance and labor market studies, Swedish National Cancer Register and other nationwide registers [10] by use of the unique personal identity number of all residents. We obtained data on PCa characteristics, co-morbidities, civil status and educational level from PCBaSe 4.1 [10]. We also collected prescribed medications data from the National Prescribed Drug Register (PDR) which was established in July 2005 [11]. The PDR contains information of all prescribed drugs dispensed at pharmacies covering the whole population of Sweden.
Moreover, information on diabetic conditions was retrieved through a linkage between PCBaSe 4.1 and the National Diabetes Registry (NDR) which was initiated in 1996 and has engaged the participation of both hospitals and primary care. This register contains detailed data on demographics, smoking, diabetes duration, treatment modalities, risk factors and diabetes complications and it currently includes most of T2DM patients age 18 and older in Sweden [12,13].
The study population included men diagnosed with T2DM, according to NDR, amongst men included in

Study population
To investigate the association of use of GnRH and a PCa diagnosis per se with diabetes control separately, we created two cohorts of men with a diagnosis of T2DM -"PCa cohort" and "PCa + GnRH cohort" (Fig. 1). In the PCa cohort, we included men with at least four registrations of data in NDR who were diagnosed with PCa after their third NDR-registration. Date of PCa diagnosis was considered as start of follow up. For each man with PCa in our study, ve PCa-free men with T2DM were randomly selected from the NDR, matched on average duration between NDR visits and number of previous NDR registrations. Start of follow-up for these men was 'inherited' from the corresponding man with a PCa diagnosis.
The PCa + GnRH cohort consisted of men with PCa and T2DM who initiated use of GnRH after the third registered date in the NDR. Date of the rst lled prescription for GnRH registered in the PDR was considered start of follow up. As a comparison, we selected ve men with PCa not on GnRH randomly from the NDR, matched on average duration between NDR visits and the number of previous NDR registrations. Start of follow up for these men was 'inherited' from the corresponding man treated with GnRH.

Exposures
The primary exposure was PCa diagnosis. In addition, we also used information on PCa risk categories and use of GnRH from NPCR and PDR. According to the National Comprehensive Cancer Network (NCCN) guideline, there are ve risk categories for PCa: 1) low-risk category: T1 or T2a stage, PSA < 10 ng/mL and Gleason score 6; ii) intermediate-risk category: T2b or T2c stage, 10 ng/mL, PSA < 20 ng/mL, or Gleason score 7; iii) high-risk category: T3a or T4 stage, PSA ≥ 20 ng/mL, or Gleason score ≥ 8; iv) regional metastases category: any T, N1 and M0 stage; v) distant metastases category: any T or N and M1 stage [14].

Outcomes
The primary outcome was worsening of diabetes control based on information collected as part of the longitudinal follow-up in the NDR. According to NICE guidance, the de nition of worsening diabetes control includes [15]: 1. HbA1c rose to 58 mmol/mol or higher (for men not already > 58 mmol/mol at baseline) 2. HbA1c was 10 mmol/mol higher than the baseline measurement. 3. Commencement of antidiabetic drugs or switch to insulin (for men not already on insulin at baseline).
We also combined the above criteria. For men whose HbA1c was less than 58mmol/L and who did not use insulin at the baseline, we included all of the above de nitions with whichever occurred rst as the combination event. For men whose HbA1c was over 58 mmol/L and/or who used insulin at the baseline, we used the remaining de nitions, with whichever came rst as the combination event.

Data analysis
We used registrations in the period 1/1 2006 to 31/12 2017. The baseline measurements for a participant were based on the three last NDR-registrations prior to the start of follow up. Missing data was imputed using last observation carried forward, i.e., if the last observation in NDR was missing then information was taken from the second last, if that was also missing it was retrieved from the third last NDR registration. If all the last three NDR observations were missing, then data was classi ed as missing.
Time to event was de ned as the time from start of follow up to the rst date of worsening diabetes control or last observation in NDR, whichever came rst. Hazard ratios (HR) and 95% con dence interval (CI) for worsening of diabetes control as de ned above were obtained using Cox proportional hazards regression models. All models were adjusted for age at PCa onset, duration of T2DM, education level, civil status, the Charlson Comorbidity Index (CCI), smoking habits, physical activity and body mass index (BMI). Cumulative incidence of worsening T2DM control was presented using Kaplan-Meier curves.   Men receiving GnRH agonists had a higher cumulative incidence for worsening diabetes control, compared to PCa free men (Fig. 2). The changes in the HbA1c measurements ( Fig. 2-a, Fig. 2-b) occurred earlier than the addition of new antidiabetic medications (Fig. 2-c).

PCa + GnRH Cohort
PCa men GnRH had a higher risk of worsening diabetes control compared to men with PCa not on GnRH agonists (HR:1.58, 95% CI: 1.39-1.80) ( Table 3). No differences by PCa risk categories were observed ( Table 3). The HR for worsening diabetes control was 1.34 (95% CI: 0.97-1.83) in regional metastatic PCa, and 1.08 (95% CI: 0.72-1.62) in men with distant metastases. a. We excluded men with a HbA1c over 58mmol/l at baseline b. We excluded men using insulin at the baseline c. This model was adjusted for age at PCa diagnosis, duration of T2DM, education level, CCI, civil status, smoking habits, physical activity and BMI Men on GnRH had a worsening diabetes control compared to men with PCa not on GnRH over time (Fig. 3). The changes in HbA1c measurements ( Fig. 3-a, Fig. 3-b) occurred earlier than the addition of new antidiabetic medications (Fig. 3-c), similar to that seen in the PCa cohort.

Discussion
In this nationwide, population-based study, use of GnRH was associated with worsening diabetes control in men with diabetes and PCa treated with GnRH.

PCa cohort
We found no statically signi cant association of PCa diagnosis (all risk categories) with worsening diabetes control in line with results of two previous studies [3,16], which investigated the effect of PCa diagnosis on glycemic control and T2DM treatments.
Advanced PCa, including regional metastatic and distance metastatic disease, was associated with worsening of diabetes control compared to men without PCa [3]. This nding may be explained by the use of GnRH in men with advanced PCa [4]. GnRH reduce insulin sensitivity [17], which could lead to worsening of diabetic control [18]. Therefore, their use may explain the association of worsening T2DM control with more advanced PCa.
In addition, starting GnRH worsened diabetes control in T2DM men compared to men with diabetes without PCa in accordance with results of previous observational studies [3,19]. These showed that men with PCa treated with GnRH agonists had an increased risk of T2DM treatment escalations, compared to men with PCa not on GnRH agonists [3] and use of GnRH increased HbA1c level and worsened diabetes control in men with PCa [19].
PCa + GnRH cohort Next, we wanted to demonstrate that the association with worsening diabetes control was caused by the GnRH rather than the PCa diagnosis per se. In the PCa + GnRH cohort T2DM men with PCa who started GnRH after PCa diagnosis had worse diabetes control than men with PCa not on GnRH, supporting the hypothesis that it is the GnRH driving the worse control.
Several biological mechanisms have been proposed to explain the association between use of GnRH and worsening diabetes control [8]. Low levels of testosterone, induced by GnRH, are implicated in the development of insulin resistance [20,21], which results in increased plasma glucose levels [22] and hence leads to worsening control of T2DM.
We found that changes in HbA1c occurred prior to changes in T2DM drug use, which is logical since antidiabetic treatment will only be changed when the deterioration of diabetes control has been veri ed on repeat measures.

Strengths and limitations
Our study has several strengths. First, we were able to assemble a nationwide population-based cohort of men with T2DM from the largest diabetes register in the world, with up to 10 years of follow up. To our knowledge, this is the largest population-based cohort study exploring the association of use of GnRH agonists with diabetes control in men with T2DM. We were also able to disentangle the impact of a PCa diagnosis itself and explore the association with different risk categories of PCa. This was achieved by selecting men with T2DM with or without PCa and also with or without GnRH separately at baseline, thereby assembling two cohorts. Secondly, we were able to look at three separate markers of worsening

Declarations
Ethics approval and consent to participate The study has been approved by The Research Ethics Board at Uppsala University, Sweden. As data of the study were obtained from the established national databases in Sweden, consent to participate and the experiment protocol were not required.

Consent for publication
Not applicable.

Availability of data and materials
The data that support the ndings of this study are available from PCBaSe Sweden, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of PCBaSe Sweden.
Funding with PCa and 28,445 PCa-free men were included in the PCa cohort. The PCa+GnRH cohort contained 692 PCa men who started GnRH after PCa diagnosis and 3,460 PCa men not using GnRH as comparison.

Figure 2
Cumulative incidence of worsening T2DM control in T2DM men by PCa status in PCa Cohort In this gure, we found that, in the PCa cohort, men receiving GnRH agonists had a higher cumulative incidence for worsening diabetes control, compared to PCa free men. The changes in the HbA1c measurements  Figure 2-(b)) occurred earlier and more obviously than the addition of new antidiabetic medications (Figure 2-(c)). When we combined the criteria to identify the event in Figure 2-(a), Figure 2-(b) and Figure 2-(c) to create the combination event, we found that cumulative incidence of combination event is higher in PCa men with GnRH compared with men without PCa (Figure 2-(d)). Figure 3