A comparison of survival and mortality outcomes for older men with localized prostate cancer treated with either primary androgen deprivation therapy or primary observation

Androgen deprivation therapy (ADT) remains the primary treatment for localized prostate cancer (PCa) even though there is no evidence that its use is beneficial in the absence of curative treatment. Finnish Cancer Registry data were utilized in this population-based study. Men aged > 70 years ( n =16534) diagnosed with localized PCa from 1985–2014, and treated either by primary observation or ADT in the absence of curative treatment, were included. The relative risk (RR) of PCa-specific mortality (PCSM) and other cause-specific mortality was determined. A life table for cause-specific survival (CSS) (i.e., PCa) and other-cause survival was created. Two age groups (70–79 years and > 80 years) and three time-based cohorts (1985–1994, 1995–2004, and 2005–2014) were compared following propensity score matching with four covariates (age, year of diagnosis, socioeconomic status, and hospital district). Follow-ups continued until death or 31 December 2015. The nationwide population-based Finnish Cancer Registry collects information on the annual incidence of cancer from hospitals, outpatient clinics, and healthcare facilities and separately from histopathological laboratories in Finland. The registry has estimated coverage of 99% for male genital cancers [(10). The complete tumor, nodes, and metastasis (TNM) classification data for all PCa cases covered by the Finnish cancer registry were retrieved in the current study. A search was conducted to identify PCa patients aged > 70 years diagnosed with localized disease (clinical stage T1/T2) from 1985–2015 and undergoing primary ADT or observation in the absence of radical treatment with curative intent. Of those identified ( n = 16534), 11572 and 4962 patients were diagnosed with PCa aged 70–79 years and > 80 years, respectively. Treatment was ADT (ADT group) and no treatment (observation group), both within four months of diagnosis. The four-month period was chosen because mandatory reports to the Finnish Cancer Registry include information about primary treatment within four months of the diagnosis. However, it is possible that patients in the observation group received later treatment with ADT and/or other palliative treatment for disease progression. There were 9704 patients in the ADT group and 6830 in the observation group (Table 1). The survival outcomes of the patients (70–79 years and > 80 years) were compared over three periods (1985–1994, 1995–2004, and 2005–2015). The Finnish Cancer Registry data were linked to the Finnish population register center database to determine the date of death. The cause of death for cancer patients was obtained from Statistics Finland.


Abstract Background
Androgen deprivation therapy (ADT) remains the primary treatment for localized prostate cancer (PCa) even though there is no evidence that its use is beneficial in the absence of curative treatment.

Material and methods
Finnish Cancer Registry data were utilized in this population-based study. Men aged > 70 years ( n =16534) diagnosed with localized PCa from 1985-2014, and treated either by primary observation or ADT in the absence of curative treatment, were included. The relative risk (RR) of PCa-specific mortality (PCSM) and other cause-specific mortality was determined. A life table for cause-specific survival (CSS) (i.e., PCa) and other-cause survival was created. Two age groups (70-79 years and > 80 years) and three time-based cohorts (1985-1994, 1995-2004, and 2005-2014) were compared following propensity score matching with four covariates (age, year of diagnosis, socioeconomic status, and hospital district). Follow-ups continued until death or 31 December 2015.
Men in their 70s undergoing observation also had better propensity score-matched 10-year CSS than those undergoing ADT.

Conclusion
Conservative management is a valid option for patients with localized PCa in whom curative treatment is not possible. 3 Background Androgen deprivation therapy (ADT) has been the cornerstone of treatment for locally advanced and metastatic (M+) prostate cancer (PCa) since the 1940s (1). Immediate ADT or ADT combined with docetaxel or with abiraterone acetate is the current treatment of choice for M+ PCa (2). However, the use of ADT increased sharply between 1989 and 2001 in the USA despite the fact that ≤ 5 % of patients with newly diagnosed PCa have distant metastases at first presentation compared with 20-25% ≥ 20 years ago (3,4). While the increased use of ADT is partly accounted for by the uptake of neoadjuvant and adjuvant treatment along with radiation therapy, it is primarily elucidated by ADT for localized disease, especially for elderly patients (5). Thus, ADT is commonly used to treat localized PCa although it has not been shown to improve survival (6).
The risk of metastases or death from conservatively managed clinical stage T1/T2 cancers was estimated in a meta-analysis of six studies from the pre-prostate-specific antigen (PSA) era (7). The risk of metastasis at 10 years was found to be 19%, 42%, and 74% for well-differentiated, moderately differentiated, and poorly differentiated tumors, respectively (7). Similarly, the long-term clinical outcomes of localized PCa without initial treatment with curative intent during the PSA era were assessed [8]. Thirty per cent of the patients died of PCa and 30% of other causes within a 12-year period (8). A landmark Swedish study demonstrated a benign course of well-or intermediately differentiated PCa in the absence of initial treatment with curative intent (9). Thus, the clinical course of localized high-risk PCa can be progressive, but the majority of cancers are indolent and slow to progress (2). According to current guidelines, observation with the option of later treatment in the case of disease progression (i.e., watchful waiting) is recommended for localized and locally advanced PCa in elderly patients with competing co-morbidities.
Thus, the current study objective was to investigate mortality in elderly PCa patients primarily treated with ADT or observation only during a long follow-up in Finland.

Study population
The nationwide population-based Finnish Cancer Registry collects information on the annual incidence of cancer from hospitals, outpatient clinics, and healthcare facilities and separately from histopathological laboratories in Finland. The registry has estimated coverage of 99% for male genital cancers [(10). The complete tumor, nodes, and metastasis (TNM) classification data for all PCa cases covered by the Finnish cancer registry were retrieved in the current study. A search was conducted to identify PCa  Table 1). The survival outcomes of the patients (70-79 years and > 80 years) were compared over three periods (1985-1994, 1995-2004, and 2005-2015 Statistical analysis CSS (PCa) and other-cause survival was evaluated using the life table method (11). The Poisson regression model was used to quantify differences in patient mortality between the defined groups. The results were reported as the relative risk (RR) of PCa-specific mortality (PCSM) and mortality from causes other than PCa. Propensity score matching was performed using four covariates (age at diagnosis, year of diagnosis, socioeconomic status/education level, and hospital district).

Results
The study population is described in Table 1 Figure 1). However, a minimal difference in CSS was observed between patients aged > 80 years undergoing either primary ADT or observation although a small difference in favor of observation was evident from 1995-2004 and 2005-2014. The results remained similar following propensity score matching (Table 3, Figure 2).
[Insert Table 3  difference was reduced following propensity score matching and was without statistical significance. Furthermore, when five-year survival from causes other than PCa was evaluated, the difference between patients undergoing observation or primary ADT was small for patients aged 70-79 years, and was further diminished after propensity score matching (Table 3, Figure 1). A significant difference in survival due to causes other than PCa over the study period was not observed in subjects aged > 80 years, and the findings remained similar following propensity score matching (Table 3, Figure 2).

Discussion
A comparison of PCa and other-cause survival and mortality in elderly male patients (aged > 70 years) with PCa was performed in this population-based study in Finland. PCSM and mortality from causes other than PCa was reduced for patients monitored with primary observation versus those managed with primary ADT. In particular, this difference was evident in patients aged 70-79 years at the time of the PCa diagnosis. CSS and survival from any cause other than PCa was also enhanced in subjects in this age group undergoing primary observation. By contrast, a smaller difference was observed between patients undergoing either ADT or primary observation with respect to PSCM, while a clear difference was not seen for other-cause mortality in elderly patients (> 80 years).
Subjects in this age group undergoing primary observation had better CSS (but not othercause survival) than those managed with primary ADT.
An increased risk of PCSM was observed in patients undergoing primary ADT, compared to those managed with primary observation, while a decrease in the risk of other-cause mortality was seen in the younger patient group with primary observation. This suggests that in recent times, primary ADT without curative treatment was generally selected for patients with aggressive disease. A 10-year three-fold risk of PCSM was associated with poorly differentiated PCa among patients with conservatively managed localized PCa in the Surveillance, Epidemiology, and End Results (SEER) Program (13). Most male patients with conservatively managed localized PCa, aged > 66 years with competing comorbidities died from causes other than PCa at 10 years, irrespective of age and tumor aggressiveness [14]. The use of primary ADT was also beneficial for patients with aggressive disease and few co-morbidities (14).
Unfortunately, information was missing on patients' co-morbidities in the current study.
However, other-cause mortality was reduced in subjects treated with observation rather than ADT. It was assumed that a treatment shift, from ADT to a more radical treatment, may have occurred for patients with good general health status. Since an increasing number of PCa cases have been diagnosed in recent years, and observation is widely utilized in Finland, it is likely that observation was selected for patients with less aggressive histology and without advanced disease. Consequently, fewer male patients died of cancer owing to the slow, natural disease course.
Survival from other-cause mortality (apart from PCa) was diminished in patients aged 70-79 years initially treated with ADT in the current study. It is still difficult to draw a conclusion from this finding owing to the absence of data on patient co-morbidities.
However, increased morbidity and mortality from ADT-related side-effects was possible.
Thus, the survival benefits of ADT are partly offset by its wide toxicity.
Many population-based analyses suggest that gonadotropin-releasing hormone (GnRH) agonist use is associated with a greater risk of coronary artery disease, myocardial infarction, and diabetes mellitus (DM) (15)(16)(17). Subsequent reports have suggested that male patients with co-morbidities or prior cardiovascular disease treated with GnRH agonists might be at increased risk of cardiovascular mortality (18,19). PSA screening practices have increased exponentially over a 30-year period, and regular PSA testing is used frequently among all socioeconomic groups in Scandinavia and Finland (26, 27). In the early years of the present study, a diagnosis of localized PCa was generally performed via a digital rectal examination or using pathological specimens obtained following a transurethral resection of the prostate. In recent years, most localized PCa cases have been diagnosed by prostate biopsies prompted by elevated PSA values. Moreover, although a proven benefit of the PSA screening of older males has not been shown, PSA testing is frequently performed for elderly patients (28). The wide-stage migration of PCa from advanced to indolent disease has been reported over ≥ 20 years (29). Thus, a commonly employed but poorly organized screening policy explains the increased rate of PCa in past decades in Finland. Consequently, the number of elderly male patients with PCa has also increased. This implies that the more favorable outcomes associated with the use of primary observation compared to ADT in patients with PCa, especially those in their 70s, can be attributed to PSA-related "lead time" rather than life extension. However, this was difficult to determine in the absence of information on PSA testing or screening of the present study cohort.
The current study had several limitations. Information was not captured on patient comorbidities and detailed PCa characteristics (i.e., Gleason scores and PSA values), so it was not possible to adjust for differences in morbidity and mortality with a potential link to ADT. Thus, study outcomes in terms of PCSM and CSS could not be compared between the different groups. Also, the absence of data on patient co-morbidities and PCa characteristics was a major study limitation. It was not possible to determine who would benefit from primary ADT in localized PCa. However, propensity score matching, particularly when applied to socioeconomic status, might have mitigated some of the comorbidity-based limitations. Detailed information on cancer treatment was also incomplete for both study treatment groups. Although men in the observation treatment group did not receive ADT or radical treatment in the four months following diagnosis, it was thought that some of them might have received treatment later. This limitation could have had a substantial effect on the study outcomes. Furthermore, although the current study population included patients with localized PCa, the reliability of staging procedures over time is debatable. TNM staging of the study population was based on mandatory reports obtained from hospitals and pathological laboratories. In other words, clinical practices in Finland were governed by national and/or European prostate cancer guidelines at the time. Consequently, the risk of metastases was evaluated according to these guidelines. It was assumed that this information was as valid as that of any registry information containing data for up to > 30 years from now.
Notwithstanding these limitations, the study had several strengths.  Test for treatment effect on prostate cancer specific mortality between age groups; p=0.083 PCa= prostate cancer, CI=confidence interval, ADT=androgen deprivation therapy.
ref=reference  Figure 1 Cancer-specific and other-cause specific survival among men between 70-79 years of age Cancer-specific and other-cause specific survival among men of 80 years and older