Trends and predictors of aggressive therapy for clinical locally advanced prostate carcinoma

Authors

  • THOMAS D. DENBERG,

    1. University of Colorado at Denver and Health Sciences Center, Denver, CO, and the University of New Mexico and New Mexico Veterans Affairs Health Care System, NM, USA
    Search for more papers by this author
  • L. MICHAEL GLODÉ,

    1. University of Colorado at Denver and Health Sciences Center, Denver, CO, and the University of New Mexico and New Mexico Veterans Affairs Health Care System, NM, USA
    Search for more papers by this author
  • JOHN F. STEINER,

    1. University of Colorado at Denver and Health Sciences Center, Denver, CO, and the University of New Mexico and New Mexico Veterans Affairs Health Care System, NM, USA
    Search for more papers by this author
  • E. DAVID CRAWFORD,

    1. University of Colorado at Denver and Health Sciences Center, Denver, CO, and the University of New Mexico and New Mexico Veterans Affairs Health Care System, NM, USA
    Search for more papers by this author
  • RICHARD M. HOFFMAN

    1. University of Colorado at Denver and Health Sciences Center, Denver, CO, and the University of New Mexico and New Mexico Veterans Affairs Health Care System, NM, USA
    Search for more papers by this author

Thomas Denberg, 4200 E. 9th Ave., B-180, Division of General Internal Medicine, University of Colorado Health Sciences Center, Denver, CO 80262, USA. e-mail: tom.denberg@uchsc.edu

Abstract

OBJECTIVE

To determine the patterns and predictors of aggressive local therapies for patients with clinically advanced (cT3) prostate carcinoma, as the USA National Cancer Institute considers external beam radiotherapy (EBRT) to be the most appropriate treatment for these patients, and currently there is less evidence supporting the use of radical prostatectomy (RP).

PATIENTS AND METHODS

We used the Surveillance, Epidemiology and End Results (SEER) cancer registries to identify patients diagnosed with cT3 disease between 1995 and 2001. Sociodemographic and clinical data included patient age, race/ethnicity, marital status, SEER registry, year of diagnosis, tumour stage and grade, and treatment. Multivariate logistic regression was used to identify significant predictors of receiving (i) RP vs EBRT, (ii) any aggressive local treatment (RP or EBRT) or no treatment.

RESULTS

Between 1995 and 2001, the proportion of men receiving aggressive local therapy for cT3 disease increased by 11% (58.4% to 69.4%), with a 20% increase in EBRT (40.3% to 60.2%) but a decline by half in RP (18.1% to 9.3%). Younger age was the strongest predictor of receiving RP rather than EBRT, and younger age with being married being a predictor of receiving aggressive local therapy (adjusted relative risk for marriage 1.33, 95% confidence interval 1.18–1.87). Black men were significantly less likely than non-Hispanic white men to receive aggressive therapy, with a relative risk of 0.56 (0.45–0.69).

CONCLUSION

By 2001, 70% of patients with cT3 disease were receiving aggressive local therapy, with EBRT 6.5 times more common than RP. Clinical trials are needed to rigorously assess the effects of different local treatment strategies on clinical outcomes in men with cT3 prostate carcinoma.

Abbreviations
(EB)RT

(external beam) radiotherapy

RP

radical prostatectomy

NCI

National Cancer Institute

SEER

Surveillance, Epidemiology and End Results

SWOG

Southwest Oncology Group.

INTRODUCTION

Initial interventions to achieve local control and possible cure for the treatment of clinically advanced (cT3) prostate cancer include external beam radiotherapy (EBRT) and radical prostatectomy (RP), either singly or combined. In general these treatments are given with neoadjuvant and/or adjuvant hormonal therapy. Recent randomized trials have shown that EBRT with adjuvant androgen-deprivation therapy significantly improved local and distant control and survival compared with EBRT alone [1–3]. This combination, endorsed by the USA National Cancer Institute (NCI), is now the most commonly recommended treatment for men with cT3 prostate carcinoma [4]. RP is not currently recommended as first-line therapy, although ongoing adjuvant chemo-hormonal trials might change this recommendation [5,6]. Thus far, observational studies suggest that RP can be beneficial for lower-risk (cT3a) disease (i.e. in men with a PSA level of <10 ng/mL, low to moderate Gleason grade, and no evidence of lymph node or seminal vesicle involvement) [4,7,8].

Little is known about clinical and sociodemographic predictors of RP and RT for the local treatment of clinically advanced prostate carcinoma. We used population-based data from the Surveillance, Epidemiology and End Results (SEER) cancer registries to characterize recent practice patterns, and to identify the clinical and sociodemographic predictors of local treatment for cT3 disease.

PATIENTS AND METHODS

We used the population-based public-use data available through the SEER cancer registries to identify men with cT3 prostate cancer. Sponsored by the NCI, the SEER registries document all incident cases of cancer for ≈ 14% of the USA population. SEER registries comprise nine geographical regions: five states (Connecticut, Hawaii, Iowa, New Mexico and Utah) and four urban areas (San Francisco-Oakland, Seattle, Atlanta and Detroit). We analysed data collected on sociodemographic characteristics, including patient age at date of diagnosis, race/ethnicity (non-Hispanic white, black, Hispanic (a US citizen or resident of Latin American or Spanish descent), and other), marital status (currently married, not married, unknown), and SEER registry where the diagnosis was made; tumour grade and stage; and therapy received within 4 months of diagnosis.

We included men with a primary diagnosis of locally advanced, lymph-node negative disease made between the years 1995 and 2001 (the most recent years available). Beginning in 1995, SEER added a field to distinguish clinical from pathological staging, replacing what had previously been a single summary field that contained information about either clinical or pathological staging if the latter had been obtained. Men were excluded if the diagnosis was made at autopsy or on the death certificate, or if data pertaining to patient age or tumour stage or grade were unavailable. Based on these criteria, there was a total of 3382 patients.

Tumour stage was based on the SEER/American Joint Commission on Cancer extent of disease (‘EOD 10 Prostate pathology ext 1995+’) variable corresponding to clinically locally advanced (cT3) disease. Tumour grade was categorized as low, moderate or high, corresponding to Gleason scores of 2–4, 5–7 and 8–10. PSA levels were not included because these are unavailable in SEER. The choice of primary local therapy was based on SEER data and was defined as prostatectomy if indicated by the site-directed surgery variable (TURP procedures were excluded), as RT if indicated by any form of EBRT or brachytherapy, and as no aggressive local therapy (generally expectant management or hormonal manipulation only) if neither surgery nor RT were reported. We defined aggressive treatment as either surgery or RT.

We evaluated the factors associated with receiving aggressive local treatment using the following predictor variables: patient age, tumour stage, tumour grade, race/ethnicity, marital status, year of diagnosis, and SEER registry. Descriptive analyses, univariate and multiple logistic regression models were used; the outcome variables were dichotomous, i.e. (i) RP vs RT and (ii) aggressive therapy (RP or RT) vs expectant management/hormonal manipulation alone. We determined a priori that the independent variables in the logistic models would be those significantly associated with initial therapy in the bivariate analysis, by using a statistical significance level of 0.25. We also examined statistical interactions of age and grade, race and grade, age and marriage, age and race, and race and marriage; none of these interaction terms was significant. Results of the logistic regression models are shown as absolute percentages of patients receiving the treatment of interest and risk ratios associated with each independent variable adjusted for the others. Risk ratios were computed as corrections of the adjusted odds ratios [9].

RESULTS

The sociodemographic and clinical characteristics of the study cohort are shown in Table 1. Within the cohort, 12.3% underwent RP, 53.2% received RT (overwhelmingly EBRT), and the remainder (34.5%) had hormonal manipulation or expectant management but no local therapy. Younger men were more likely to have RP, with almost 30% of men aged <55 years having RP compared with only 2% of men aged ≥ 80 years.

Table 1. 
Sociodemographic and clinical characteristics of 3382 men with clinically staged, locally advanced prostate carcinoma
CohortAll menTreatment
RPEBRTWW/HT
  1. Treatment refers to the % of men who received as primary local therapy RP, any form of EBRT or brachytherapy, or systemic therapy only, consisting of conservative management (WW) or hormonal manipulation (HT). Married refers to the % of men currently married; Gleason Grade corresponds to Gleason score (well-differentiated = 2–4; moderately differentiated = 5–7; poor or undifferentiated = 8–10). †Not otherwise specified.

Number of patients338241517991168
Age category in years, %
 <558.329.651.319.0
 55–599.122.755.921.5
 60–6415.520.955.521.5
 65–6922.313.961.524.6
 70–7423.38.956.629.6
 75–7916.43.854.042.2
 ≥805.12.023.274.8
Gleason grade, %
 2–41.814.558.227.3
 5–755.113.758.527.8
 8–1043.011.350.138.6
Clinical stage, %
 T3a50.712.356.531.2
 T3b (seminal vesicle)34.013.652.933.5
 T3 (periprostatic, NOS)15.39.745.145.2
Race/ethnicity, %
 White78.813.054.832.1
 Black10.311.647.940.5
 Hispanic4.07.256.536.2
 Other6.99.746.543.9
Marital status, %
 Married77.413.656.629.8
 Not22.610.748.241.1
SEER registry, %
 San Francisco/Oakland24.411.054.434.6
 Connecticut8.512.748.938.4
 Detroit14.413.557.029.6
 Hawaii4.56.547.346.2
 Iowa12.311.753.235.1
 New Mexico4.112.253.234.5
 Seattle17.914.248.437.3
 Utah6.36.158.935.1
 Atlanta7.620.860.418.8

In 2001, the most recent year for which data were available, and among men aged <70 years (the group most likely to receive some form of aggressive therapy), 16.3% had RP and 23.2% had no therapy for local control. Compared with non-Hispanic white men, black men had similar rates of RP (13.0% vs 11.6%) but were less likely to have local treatment in general (RT or RP; 59.5% vs 67.8%, P = 0.01). Married and unmarried men had similar rates of RP, but married men were more likely to receive any form of local therapy (70.2% vs 58.9%, P < 0.001). Finally, rates of RP among patients with seminal vesicle involvement and high-grade tumours were 13.6% and 11.3%, respectively.

Figure 1 shows the temporal trends in therapy for cT3 prostate carcinoma over a 6-year period between 1995 and 2001. Rates of aggressive local treatment (RP or RT) increased by 11% (58.4% to 69.4%). This corresponded with a halving in the frequency of RP (18.1% to 9.3%) and a 20% absolute increase in RT (40.3% to 60.2%). The period between 1995 and 1998 represented that with the most rapid change, after which treatment patterns stabilized.

Figure 1.

Therapeutic trends for local control of cT3 prostate carcinoma.

The results of the multivariate analyses, shown in Tables 2 and3, give the adjusted risk ratios for receiving RP vs RT vs any local therapy (RP or RT) vs only hormonal manipulation or expectant management. Younger age was the strongest predictor of RP over RT, and any form of aggressive treatment vs none. Black and unmarried men had a significantly lower relative risk of receiving any local therapy than non-Hispanic whites and married men, respectively. Men with high-grade tumours were significantly less likely than men with lower-grade tumours to receive local therapy. Finally, there was some regional variability, with having any local therapy less frequent in San Francisco/Oakland, Connecticut and Iowa than in the other SEER registries.

Table 2. 
A summary of logistic regression predicting RP vs RT for local control of cT3 prostate carcinoma
Variable/modelOverall relative risk (95% CI)
  • *

    Variables that retained significance in stepwise logistic regression; original predictor variables included patient age, tumour stage, tumour grade, race/ethnicity, marital status, year of diagnosis, and SEER registry.

Total number2214
Age category, years
 <55   1.34 (1.24–1.41)*
 55–59   1.18 (1.09–1.25)*
 60–64   1.16 (1.08–1.23)*
 65–69 (reference)
 70–74   0.94 (0.87–0.99)*
 75–79   0.89 (0.80–0.95)*
 ≥80   0.91 (0.77–1.00)
Year of diagnosis
 1995 (reference)
 1996   0.97 (0.85–1.07)
 1997   0.83 (0.72–0.93)*
 1998   0.76 (0.64–0.86)*
 1999   0.73 (0.63–0.84)*
 2000   0.75 (0.63–0.86)*
 2001   0.75 (0.62–0.86)*
SEER registry
 San Francisco/Oakland   0.92 (0.82–0.99)*
 Connecticut   1.01 (0.90–1.10)
 Metro Detroit (reference)
 Hawaii   0.92 (0.76–1.01)
 Iowa   0.94 (0.84–1.03)
 New Mexico   0.97 (0.81–1.08)
 Seattle   1.01 (0.92–1.09)
 Utah   0.92 (0.80–1.00)
 Metro Atlanta   1.05 (0.92–1.15)
Table 3. 
A summary of logistic regression predicting any aggressive (RP or RT) vs no local control therapy for cT3 prostate carcinoma
Variable/modelOverall relative risk (95% CI)
  • *

    Variables that retained significance in stepwise logistic regression; original predictor variables included patient age, tumour stage, tumour grade, race/ethnicity, marital status, year of diagnosis, and SEER registry.

  • †Not otherwise specified.

Total number3382
Age category, years
 <55   1.42 (1.02–1.91)*
 55–59   1.25 (0.92–2.09)
 60–64   1.24 (0.97–1.56)
 65–69 (reference)
 70–74   0.80 (0.65–0.97)*
 75–79   0.59 (0.48–0.71)*
 ≥80   0.33 (0.26–0.41)*
Race/ethnicity
 White (reference)
 Black   0.56 (0.45–0.69)*
 Hispanic   0.75 (0.54–1.01)
 Other   0.81 (0.62–1.01)
Married   1.33 (1.18–1.87)*
Gleason grade
 2–4   0.97 (0.56–1.53)
 5–7 (reference)
 8–10   0.73 (0.66–0.90)*
Clinical stage
 T3a (reference)
 T3b (seminal vesicles)   0.89 (0.78–1.02)
 T3 (periprostatic, NOS)   0.74 (0.63–0.86)*
SEER registry
 San Francisco/Oakland   0.79 (0.62–0.98)*
 Connecticut   0.44 (0.30–0.63)*
 Metro Detroit (reference)
 Hawaii   0.78 (0.54–1.03)
 Iowa   0.69 (0.53–0.89)*
 New Mexico   0.74 (0.49–1.04)
 Seattle   0.82 (0.65–1.02)
 Utah   0.77 (0.55–1.04)
 Metro Atlanta   1.23 (0.83–1.76)

DISCUSSION

Population-based SEER data showed that the proportion of men receiving local therapy for locally advanced prostate cancer increased by 11% (58% to 69%) from 1995 to 2001. During this period, particularly between 1995 and 1998, fewer men had RP while a substantially higher proportion had RT. Younger age was a strong predictor of having both any local therapy and RP, specifically. Older men, black men, unmarried men and men with higher-grade tumours were significantly less likely than their counterparts to have any local therapy. These treatment patterns are very similar to those of clinically localized (cT1/T2) disease, in which significant predictors have included tumour stage and grade, SEER registry, and patient age, race/ethnicity and marital status [10,11].

Almost a quarter of men aged <70 years in the 2001 had no local therapy for their cT3 prostate cancer. This is potentially of concern, although we do not know what proportion of these men had significant comorbidities, were unwilling to accept local therapy when offered, received delayed local treatment, or did not receive recommendations for local therapy from their urologists or radiation oncologists.

Because RP is the most aggressive form of therapy it is often considered the most definitive. This impression might explain why it is sometimes favoured in younger men whose anticipated life-span is greatest. ‘Getting the tumour out’ and using surgical pathology to ‘know what one is dealing with’ might be preferred over allowing the tumour to remain within the body without knowing the pathological stage, as happens with RT. When RP is used for local control, surgical staging has been shown to be clinically valuable. Recent results from the randomized multicentre Southwest Oncology Group (SWOG) study 8794 indicate that immediate adjuvant RT after RP in the setting of surgically staged T3 carcinoma is better than a ‘wait-and-see’ approach after RP [12]. However, it is unclear for both younger and older patients whether the clinical outcomes associated with this approach are better than those that could be obtained with RT and androgen deprivation only.

The proportion of men with cT3 disease who had RP halved from 1995 to 2001. The future use of RP is uncertain. One small phase II study (SWOG 9109) showed favourable outcomes associated with neoadjuvant antiandrogen therapy before RP for clinical T3/T4 prostate carcinoma [13]. However, at least three randomized trials evaluating 3 months of neoadjuvant androgen deprivation before RP in patients with T1–T3 disease failed to show better outcomes for cancer control [14–16]. Two ongoing adjuvant chemo-hormonal trials after RP (SWOG 9921 and Veterans’ Affairs Cooperative Study ♯553) [5,6] might show that patients with high-risk characteristics have very favourable outcomes.

A recent case series reported comparable 15-year cancer control and survival outcomes with RP for cT3 and cT2 disease [17]. However, in that retrospective study the mean preoperative PSA level of patients with cT3 disease was only 10.2 ng/mL, and the proportion of patients with clinical seminal vesicle involvement was not reported. In addition, the study was limited to cases within a single tertiary referral centre, and there was significant heterogeneity in the receipt and timing of neoadjuvant hormonal therapy, as well as adjuvant and salvage hormonal therapy and RT.

Only one very small randomized trial compared RP against EBRT using hormonal ablation both before and after primary therapy in stage T2c and T3 prostate carcinoma [18]. That study showed no difference in 5-year survival, but higher levels of late morbidity and a lower quality of life in the RP arm. However, the long-term adverse effects of RT remain to be assessed. A recent retrospective analysis of a large group of patients found a 1.7-fold increase in rectal cancer compared to patients having surgery only [19], although another epidemiological study found no such effect [20]. Considerations such as these might lead some urologists to counsel their patients on what they perceive are advantages of surgery, especially in younger men.

The finding that black and unmarried men were less likely to receive local therapy might be attributable to higher levels of comorbidity in these groups. Unfortunately, comorbidities were not recorded in this dataset. However, in a separate study of treatment patterns for T1/T2 disease, married men with no known comorbidities were significantly more likely to receive aggressive therapies than a similar cohort of unmarried men [10]. Possibly, married men or their wives advocate therapy that they perceive as most likely to give a cure, whereas unmarried men are less likely to have social supports encouraging aggressive interventions. Clinicians might also recommend aggressive therapies more strongly or more often to married than unmarried men. The perceived ability of wives to help care for patients during and after treatment might play a role. The influence of spouses on treatment decision-making deserves greater scrutiny, as it appears to be pervasive in treatment choices for other stages of prostate cancer [10].

On balance, studies evaluating RT have included more patients than those evaluating RP. To determine the optimum treatment of cT3 disease, larger trials of RP are needed and, ideally, there should be trials that directly compare surgery and RT as parts of multimodal treatment regimens that incorporate hormonal therapy and/or chemotherapy. Clinical downstaging is possible with RP, but because upstaging is more likely these trials should also consider how often RP results in the need for salvage RT.

The present study of treatment patterns has several limitations. First, regardless of whether a patient received therapy for local control, we assumed that all patients received hormonal manipulation, because this is a standard of care. However, information on hormonal therapy was unavailable in the dataset, so possibly a proportion of cases had no systemic treatment. In addition, SEER only captures primary therapy received within 4 months of diagnosis. Therapy provided after this period would have resulted in an ascertainment bias that could, e.g. have differentially affected black men, unmarried men, older men, or any other particular group. Information about PSA levels, comorbidities, and socio-economic status was also lacking; including this information would have made the analyses more robust. Finally, because comorbidities were not recorded in the dataset, results are suggestive of, but inconclusive, for the presence of disparities by race/ethnicity and marital status in receipt of any aggressive local treatment. Despite these limitations we showed that rates of local-control interventions for cT3 prostate carcinoma increased over a recent 6-year period, and that these changes corresponded with declining rates of RP and increasing rates of RT.

ACKNOWLEDGEMENTS

This study was supported in part by a Redes en Acción research subcontract from the Baylor College of Medicine, funded by a Special Populations Network NCI grant (U01CA86117), Amelie G. Ramirez, DrPH (PI). This study used the SEER database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the Applied Research Program, NCI and the SEER Program tumour registries in the creation of this database.

CONFLICT OF INTEREST

None declared. Source of funding: NCI U01CA86117.

Ancillary