Relative contribution of androgen deprivation therapy in combination with adjuvant radiotherapy for node-positive prostate cancer.

Abstract

Background

In men with node-positive prostate cancer after radical prostatectomy, the value of androgen deprivation therapy (ADT) in combination with adjuvant radiotherapy has not been established.

Objective

To examine the oncologic benefit of ADT in combination with adjuvant radiotherapy for node-positive prostate cancer.

Methods

We analyzed data for 372 prostate cancer patients treated at San Raffaele Hospital with adjuvant radiotherapy for node-positive disease after radical prostatectomy, 272 received both ADT and radiotherapy. Eight-six men were followed without an event for more than 10 years.

Results

Patients who received adjuvant radiotherapy + ADT had more aggressive disease, with higher preoperative PSA level, higher rate of ISUP grade 5, pT3b-T4 tumors and ≥3 positive nodes. At multivariable Cox regression, there was no significant difference in overall survival between men treated by adjuvant radiotherapy + ADT vs radiotherapy alone (hazards ratio [HR]: 0.91; 95% confidence interval [CI]: 0.45, 1.84; p=0.8). Men treated by adjuvant radiotherapy + ADT had a non-significantly higher risk of prostate cancer death than patients who received adjuvant radiotherapy alone (HR: 5.39; 95%CI 0.70, 41.39; p=0.11). These results remained consistent in a number of sensitivity analyses, including propensity score matching.

Conclusions

We can exclude a large benefit to ADT on prostate cancer death in men with node-positive disease receiving adjuvant radiotherapy. Men should not be offered ADT in this setting outside well-controlled clinical trials.

INTRODUCTION

Lymph node metastases are found in approximately 15% of patients treated by radical prostatectomy. Although nodal involvement is clearly a poor prognostic sign, it is not an inevitable harbinger of recurrence: about one in three patients with positive lymph nodes remain recurrence free at long-term follow-up even in the absence of post-operative treatment¹. This suggest prognostic heterogeneity in men with positive lymph nodes that should be taken into consideration in order to avoid overtreatment.

Androgen deprivation therapy (ADT) is the gold standard treatment for metastatic prostate cancer. Historically, patients with nodal metastases after radical prostatectomy were managed with ADT on the grounds of improved survival over postoperative observation². Subsequent evidence that a combination of ADT and adjuvant radiotherapy is beneficial over ADT alone³ has contributed to a shift in the treatment paradigm, with positive lymph nodes no longer considered a sign of disseminated disease but potentially cured by treatments aimed at local control. The rationale behind the use of radiotherapy in this setting is that nodal metastases are deemed adverse pathologic feature after radical prostatectomy, expression of advanced, high-risk (but localized) disease. Given this premise, it seems reasonable that cancer control might result from local rather than systemic therapy, in keeping with overwhelming evidence that adjuvant radiotherapy is beneficial in node-negative patients with other adverse pathologic features^4,5. This raises the obvious question of the relative contribution of ADT to treatment effect. Research data are sparse on this point, having focusing predominately on whether radiotherapy adds to ADT rather than the other way around^6,7.

ADT causes a large number of side-effects, including hot flashes, fatigue and impaired libido, as well as metabolic and cardiovascular complications⁸. Moreover, there is evidence that ADT increases the risk cardiovascular death³. For these reasons, the use of ADT in combination with adjuvant radiotherapy for node-positive prostate cancer would only be justified if it resulted in a large decrease in cancer-specific death.

To test whether the addition of ADT to adjuvant radiotherapy improves survival compared to radiotherapy alone, we examine a cohort of node-positive prostate cancer patients treated at high-volume institution.

METHODS

We analyzed data of 643 prostate cancer patients who received adjuvant radiotherapy with or without ADT for node-positive disease after radical prostatectomy. All patients received surgery as primary treatment at San Raffaele Hospital between 1991 and 2017. An extended pelvic lymph node dissection was performed in all the cases, which included the removal of obturator, external iliac, and hypogastric nodes. Treatment was considered adjuvant if it was administered within six months from surgery and all patients were followed for more than six months. The decision to administer additional ADT was based on the clinical judgement of each treating physician according to individual patient and cancer characteristics. We excluded patients who received neoadjuvant treatment (n=229) and those who had missing pathologic data (n=42), resulting in 372 men eligible for the analyses.

Adjuvant radiotherapy consisted of local radiation to the prostatic bed with or without the seminal vesicle bed and pelvic lymph nodes area (whole-pelvis radiotherapy). All patients were treated with high-energy photon beams (6–18 mV) at conventional fractionation (1.8–2 Gy/ fraction), at a median dose of 68 Gy (interquartile range [IQR]: 66, 70) using previously described techniques^3,9. A full description of the methods of radiotherapy is available in Supplementary appendix A.

Adjuvant androgen deprivation therapy consisted of either bilateral orchiectomy or luteinizing hormone releasing hormone agonist. Androgen deprivation therapy was generally intended to be lifelong. However, given the retrospective nature of our study, it is uncertain whether patients discontinued treatment after a period of androgen deprivation therapy.

The primary outcome of the study was cancer-specific survival. Secondary outcomes were overall survival and clinical recurrence (CR), defined as positive imaging in case of PSA ≥ 0.2 ng/ml in two consecutive measurements. The cause of death was defined by the attending urologist or oncologist who followed the patients or by death certificate.

Our statistical analyses consisted of several steps. First, we compared disease characteristics between the groups. Differences were examined using the Wilcoxon rank-sum and Chi-squared tests. Second, Kaplan-Meier methods were used to estimate cancer-specific and overall survival in the two groups. Cox proportional hazards regression was used to compare survival between the groups. The Cox model was adjusted for age, number of positive nodes (categorized as 1–2 vs 3+) and the risk of biochemical recurrence derived from the MSKCC nomogram¹⁰, which includes preoperative (PSA level) and pathologic (grade, stage, nodal involvement and margins status) variables. Third, we built a competing risk regression model with cancer-specific mortality as the outcome and death from other causes as the competing event. Finally, we conducted a number of sensitivity analysis to assess the robustness of our findings. To determine whether differences in outcome were related to baseline differences between groups, we restricted our analysis to men with pathologic ISUP grade 3 to 5. In a separate analysis, we also excluded patients treated before 2005, when a different ISUP grading system was used. To test whether the effect of ADT might differ according to nodal burden, we repeated the analyses in patients with 1–2 vs 3+ positive nodes¹¹. Finally, since the probability of being treated by radiotherapy alone or in combination with ADT may be affected by disease characteristics, we used a propensity score approach. The individual probability of receiving adjuvant radiotherapy + ADT was calculated using a logistic regression model according to age, MSKCC nomogram-derived risk of BCR and number of positive nodes. This likelihood was then used to match in a 1:2 ratio patients in the radiotherapy group to men treated by radiotherapy + ADT with similar (−/+ 5%) probability in order to create a more homogeneous subcohort.

RESULTS

Table 1 describes the characteristics of our study cohort. Although differences between the groups were statistically significant only in few cases, men receiving adjuvant radiotherapy + ADT had more aggressive disease, namely higher preoperative PSA level, higher rate of ISUP grade 5, pT3b-T4 tumors and greater metastatic burden (≥3 positive nodes).

Table 1.

Descriptive characteristics of 372 patients treated by radical prostatectomy who had N+ disease at surgical pathology, stratified by adjuvant treatments.

	Adjuvant Radiotherapy (N=100; 27%)	Adjuvant Radiotherapy + ADT (N=272; 73%)	P value
Year of surgery
<2005	17 (17%)	98 (36%)	0.002
2006–2010	39 (39%)	90 (33%)
>2010	44 (44%)	84 (31%)
Age, years	67 (60, 71)	64 (60, 69)	0.077
Preoperative PSA level, ng/ml	9.4 (6.6, 19.0)	11.1 (7.3, 19.5)	0.2
Pathologic ISUP grade
1–2	22 (22%)	44 (16%)	0.5
3	26 (26%)	69 (25%)
4	14 (14%)	34 (13%)
5	38 (38%)	125 (46%)
Pathologic stage
T2–T3a	39 (39%)	83 (31%)	0.12
T3b–T4	61 (61%)	189 (69%)
Positive surgical margins	46 (46%)	157 (58%)	0.044
Number of positive nodes
1–2	82 (82%)	172 (63%)	0.001
3+	18 (18%)	100 (37%)

There were 48 all-cause and 18 cancer-specific deaths. Median follow-up for survivors was 77 months (IQR: 44, 113) with 86 patients followed for more than 10 years without an event. The predicted 10-year overall survival was 81% (95% confidence interval [CI]: 63%, 91%) for patients treated by radiotherapy alone and 85% (95%CI: 78%, 90%) for those who received radiotherapy and ADT (Fig. 1a). The predicted 10-year cancer-specific survival was 98% (95%CI: 87%, 100%) for the radiotherapy group and 92% (95%CI: 87%, 95%) for men who received radiotherapy + ADT (Fig. 1b).

Figure 1.

Kaplan Meier curves for overall (A) and cancer-specific (B) survival stratified by treatment group. aRT: adjuvant radiotherapy. ADT: androgen deprivation therapy

The results of our Cox regression analyses are shown in Table 2. There was no significant difference in overall survival between men treated by adjuvant radiotherapy + ADT vs radiotherapy alone (hazards ratio [HR]: 0.91; 95%CI: 0.45, 1.84; p=0.8). Men treated by adjuvant radiotherapy + ADT had a non-significantly higher risk of prostate cancer death than patients who received adjuvant radiotherapy alone (HR: 5.39; 95%CI 0.70, 41.39; p=0.11). This finding was confirmed in the competing risk analysis used to predict cancer-specific death with death from other causes as the competing event, with the HR being 5.60 for comparing men treated with adjuvant radiotherapy + ADT vs adjuvant radiotherapy alone (95%CI 0.68, 45.86; p=0.11). Figure 1 shows differences between groups in overall and cancer-specific survival. These results are compatible with a limited oncologic benefit of androgen deprivation therapy when administered in combination with adjuvant radiotherapy for node-positive prostate cancer.

Table 2.

Multivariable Cox proportional hazard model to assess the association between adjuvant treatments and oncologic outcomes.

Variable	Hazards ratio	95% confidence interval	p-value
All-cause mortality
Adjuvant treatment
Radiotherapy	Ref
Radiotherapy + ADT	0.91	0.45, 1.84	0.8
Cancer-specific mortality
Adjuvant treatment
Radiotherapy	Ref
Radiotherapy + ADT	5.39	0.70, 41.39	0.11
Other-cause mortality
Adjuvant treatment
Radiotherapy	Ref
Radiotherapy + ADT	0.49	0.22, 1.09	0.080

Models adjusted for age, risk of BCR according to the MSKCC nomogram and number of positive nodes (1–2 vs 3+)

Abbreviations: ADT: androgen deprivation therapy.

A total of 77 patients developed clinical recurrence. The predicted 10-year clinical recurrence-free survival was 92% (95%CI: 82%, 96%) for patients treated by radiotherapy alone and 70% (95%CI: 63%, 76%) for those who received radiotherapy and ADT. At multivariable analyses, patients who received radiotherapy and ADT had significantly higher risk of clinical recurrence (HR: 2.41; 95%CI: 1.09, 5.31; p=0.029; Table 2).

Results of our sensitivity analyses are described in supplementary appendix B. In brief, our findings were unaltered when the analyses were restricted to patients with ISUP grade 3 to 5 tumors and to the subgroup treated after 2005. We did not find evidence of a different effect of ADT according to the number of positive nodes: the risk of clinical recurrence in the ADT group did not differ in case of positivity in 1–2 (HR: 2.05, 95%CI: 0.79, 5.33; p=0.14) vs 3+ (HR: 3.93, 95%CI: 0.93, 16.65; p=0.063) nodes. Although the probability of receiving ADT was similar between the two groups (supplementary appendix C), our results were unaffected in the subcohort of patients selected after propensity score matching with cancer-specific survival being non-significantly poorer on radiotherapy + ADT vs radiotherapy alone (HR: 3.27; 95%CI: 0.37; 28.46; p=0.3). Results were also similar for clinical recurrence (HR: 2.18; 95%CI: 0.88, 5.41; p=0.094).

DISCUSSION

We were not able to demonstrate a survival benefit from ADT administered with adjuvant radiotherapy in men with node-positive prostate cancer. Moreover, the lower bound of the 95% C.I. for the effect of ADT on cancer-specific survival excluded a large benefit.

In modern medicine, the risk/benefit ratio is key for treatment decisions: a patient would be likely unwilling to undergo radical prostatectomy if it only reduced quality of life without curing cancer. Similarly, the potentially life-threatening side effects of ADT⁸ appear acceptable only if its administration entails a compelling oncologic benefit. In this regard, previous evidence showed improved survival for ADT over postoperative observation¹², but a prognostic advantage over other treatment strategies such as radiotherapy has never been observed. Rather, prior investigators comparing postoperative treatments for node-positive disease found that the risk of metastases was higher for patients treated by ADT alone than adjuvant radiotherapy alone¹³. In a similar study, the 5-yr overall survival probability was lower for men receiving ADT alone than for those treated by radiotherapy alone and postoperative observation¹⁴. This data suggests that if an intervention is needed in case of nodal involvement at surgical pathology, that should be radiotherapy rather than ADT. This is also consistent with other papers showing improved survival for ADT + radiotherapy than ADT alone^3,15,16. On the other hand, evidence supporting the addition of ADT in patients treated with radiotherapy is quite limited. In a population-based study, Wong et al examined different management strategies for pN+ patients: at pairwise analysis, radiotherapy + ADT was not superior than radiotherapy alone for overall survival¹⁴. By contrast, the comparison with observation at multivariable analysis showed that patients treated with radiotherapy + ADT had better survival, while those who received radiotherapy alone did not¹⁴. However, although it is challenging to isolate the oncologic effect of a treatment using overall survival, the counterintuitive effect of certain covariates (e.g. higher grade improved survival) seems compatible with the presence of unmeasured confounders. The relative contribution of hormonal therapy in this setting may also be extrapolated from a subgroup analysis of the RTOG 85–31 trial. Therein, investigators assessed the impact of ADT in N+ patients undergoing radiotherapy: in a subcohort of 42 patients treated after radical prostatectomy, the study did not show oncologic benefit associated with hormonal therapy¹⁷. Thus, the added value of a combination strategy over radiotherapy alone is far from established. Our series provides important insight on this issue as we found that the addition of ADT to radiotherapy did not improve oncologic outcomes. More importantly, the confidence interval for cancer-specific survival suggests that the oncologic benefit necessary to justify the morbidity of ADT is extremely unlikely, and this is also supported by similar results for clinical recurrence. At the same time, since ADT is associated with an increased risk of death from other causes³, our results suggest that ADT should not be given in this setting outside well-controlled clinical trials.

Our study is not devoid of limitations. For example, we cannot rule out residual confounding from known or unknown variables. Since the study included patients treated over more than 25 years, it is possible that aspects of clinical care (e.g. surgical technique, indication for lymphadenectomy) have changed over time. Although this is problematic in every observational study, our results remained consistent in a number of sensitivity analyses. In addition, consideration of baseline characteristics might raise concerns for selection bias, that is, patients treated by radiotherapy and ADT had more aggressive disease. To address this issue, we performed propensity score matching with no meaningful differences in survival results. Moreover, it is noteworthy that a similar concern has been raised in a prior paper³. However, patients treated with radiotherapy + ADT had better overall and cancer-specific survival than those who received ADT alone despite worse prognostic profile and as such, the added value of radiotherapy was claimed. In our study, although men treated by radiotherapy and ADT had similarly more aggressive disease, we did not observe a survival difference. This seems more compatible with a limited contribution of ADT than with a selection bias.

Our results argue against current guidelines that recommend a combination of ADT and radiotherapy as treatment option for node-positive disease^6,7. In this regard, our findings support the inclusion of radiotherapy alone among postoperative strategies as the addition of ADT was not of added value. Historically, identification of nodal metastases during radical prostatectomy was an indication for discontinuing surgery. The belief that nodal metastases were a sign of systemic cancer was the rationale for the use of ADT, the standard of care for metastatic prostate cancer. Consideration of nodal involvement has changed, and nowadays it is unlikely that a patient would be told he has systemic disease in case of N+ pathology. By contrast, the administration of ADT seems guided by cultural inheritance, that is, evidence of metastases mandates systemic therapy. However, it has been established that patients with nodal metastases have remarkably better prognosis that those who have bone or visceral metastases^18,19, suggesting that these are biologically different phases of tumor spread. For this reason, it is plausible that ADT might not be as effective in nodal metastases as it is for systemic disease. In addition, the fact that radiation and ADT are both effective treatments for node-positive patients does not imply they should be administered together. If we apply the same concept to newly diagnosed, clinically significant prostate cancer, a combination of surgery and radiotherapy should always be considered as each treatment is more effective than observation. However, overwhelming evidence proved that a combination of treatments is beneficial only for specific subgroups of patients in whom additional side effects are therefore justified. By the same token, ADT should be properly investigated (i.e. using the adequate reference group) to get insight on the ratio between the benefit of this treatment and its side effects.

Our results have implications for empirical research. If ADT does not invariably improve survival in node-positive patients treated with adjuvant radiotherapy, a survival benefit for particular subgroups of patients is still plausible, as observed for adjuvant radiotherapy²⁰. In the last decade we have been asking who might benefit from local control among pN+ patients: our findings suggest that such question should be rather who may benefit from systemic therapy. Thus, a challenge for empirical research is to identify patients in whom the side effects entailed by ADT might be justified by a survival benefit. We intend to investigate such possibility in future studies.