Evolving Landscape of Practice Patterns in the Management of Localized Low-Risk Prostate Cancer: A NCDB Study

Abstract

Introduction: 

Most patients with low-risk prostate cancer (LRPC) are candidates for active surveillance (AS) or definitive local treatment, including radical prostatectomy (RP), external beam radiotherapy (EBRT), or prostate seed implant (PSI). In this study, we analyzed the treatment pattern of patients with LRPC using a large patient population provided by the National Cancer Database (NCDB).

Methods: 

Patients with LRPC, as defined by the National Comprehensive Cancer Network criteria, between 2004 and 2015 from the NCDB were analyzed. Patients, with no or minimal comorbidity, were categorized according to treatment modalities, including RP, EBRT, PSI, and no local therapy (NLT). Chi-square tests were used to analyze the association among demographic variables and treatments. Multivariate logistic regression analyses were conducted to determine the likelihood of choosing NLT.

Results: 

In total, 195,452 patients diagnosed with LRPC between 2004 and 2015 were identified. There was a significant change in practice patterns with increased utilization of NLT over time, from 11.3% in 2004 to 53.5% in 2015, with a turning point in 2009. RP utilization peaked at 41.6% in 2008, EBRT peaked at 24.3% in 2006, and PSI peaked at 35.3% in 2004. However, it dropped to 17.6%, 18.1%, and 10.8%, respectively, in 2015. Academic/research programs (A/RP), older age, lower income, poorer education, lack of insurance, African Americans, and other races were associated with increased NLT use. The difference between White and African American, and older and younger patients in odds ratios were less after 2010.

Conclusions: 

There was a significant shift in treatment patterns between 2004 and 2015. AS, as represented by the NLT in this study, has become the dominant treatment of choice for patients with LRPC. A/RPs were leading this trend, particularly in younger patients. Socioeconomic factors played roles in choosing NLT, but the racial difference became less apparent after 2010.

Introduction

As the most prevalent cancer and the second leading cause of cancer mortality in American men, prostate cancer is a heterogeneous disease with a wide spectrum of clinical manifestations owing to underlying genetic and epigenetic alterations.¹ In clinical practice, it is subcategorized into low-risk, intermediate-risk, and high-risk groups based on pretreatment prostate-specific antigen (PSA) levels, the amount of cancer detected at biopsy, and Gleason score as proposed by D'Amico et al.²

Most patients with localized, low-risk prostate cancer (LRPC) are eligible for either active surveillance (AS)/watchful waiting (WW) or definitive local treatment, including radical prostatectomy (RP), external-beam radiation therapy (EBRT), or prostate seed implant (PSI). Among these treatment modalities, there is currently no single best option regarding oncological outcomes or adverse effects. Treatment decisions are frequently influenced by patient characteristics, such as age, performance status, comorbidities, and physician and/or patient preferences.

In recent years, there has been a growing trend favoring AS in the management of LRPC worldwide and across racial and age groups.^3,4 Several studies on the AS approach have demonstrated that it is safe and feasible in a timeframe of 5 to 10 years,^5-9 especially for a fraction of LRPC, those with very low-risk prostate cancer, as defined by Epstein et al.¹⁰ Despite these advances, management decisions for localized LRPC remain a source of contention. Recently, the ProtecT trial included a patient population in the modern PSA screening era, with a significant proportion of patients in the LRPC category. After a median of 10 years of follow-up, the prostate-specific mortality was low, with no difference between intervention groups, although the AS group developed more metastases than the RP or radiation groups.¹¹ However, the relatively small sample size in each cohort and fewer events would necessitate a much longer follow-up period, most likely decades, to see the outcomes.

In this study, we endeavored to take advantage of the large number of patients with prostate cancer obtained from the National Cancer Data Base (NCDB). We examined all patients with LRPC in the NCDB between 2004 and 2015 for the practice patterns and treatment modalities used. To ensure patient comparability across the treatment cohorts, we only included patients with a Charlson-Deyo comorbidity score of 0 or 1, thus reducing the likelihood that treatment decisions would be influenced by comorbidity.

Methods

Patient Selection

The NCDB is a comprehensive hospital-based clinical oncology database supported by American College of Surgeons' Commission on Cancer (CoC) and the American Cancer Society. The NCDB collects data from over 1500 CoC-accredited hospitals and accounts for approximately 70% of newly diagnosed cancer cases in the United States.¹² Because the patient data were deidentified, they were exempt from oversight by the institutional review board. We analyzed the NCDB between 2004 and 2015 for patients with low-risk prostate cancer based on the D'Amico and the National Comprehensive Cancer Network (NCCN) classification (PSA <10 ng/ml, cT1-cT2a, and Gleason score ≤6).^13,14 Patient and clinical characteristics were collected to examine trends in patient care. Patients with non-LRPC diagnoses, missing clinical staging data, and missing survival data were excluded from this study. The Charlson-Deyo comorbidity score is commonly used to predict mortality based on chronic comorbidities. Only patients with Charlson-Deyo comorbidity scores of 0 or 1 were included to ensure that the cohorts were comparable. The Consolidated Standards of Reporting Trials diagram illustrated in Figure 1 demonstrates the exclusion criteria used to define the cohort. A total of 195,452 patients who met the eligibility criteria were included in this study.

Figure 1.: CONSORT diagram. CONSORT, Consolidated Standards of Reporting Trials; GS, Gleason score; PCa, prostate cancer.

Patient and Clinical Characteristics

The following patient characteristics were obtained: age at diagnosis (years), race (White, Black, or other), insurance status (private, Medicaid/other government insurance, Medicare, or uninsured), median income (≤$30,000, $30,000-35,999, $36,000-45,999, or ≥$46,000), and education level (percentage of patients without a high school diploma: ≥21%, 13-20.9%, 7-12.9%, or <7%). The median income and educational level were calculated using the patient's zip code. Charlson-Deyo score (0, 1, ≥2), Gleason score, PSA level, tumor, node, metastasis (TNM) stage, and treatment modalities were obtained.

Treatment

The mode of treatment was the most significant independent variable. Patients were classified based on their treatment, which included local therapies, such as RP, EBRT, PSI, and no local treatment (NLT). Owing to the limitations of the NCDB, it is impossible to distinguish between AS, WW, and other reasons for patients' refusal to receive local treatment. Patients who received palliative care, isolated hormone therapy, or any treatment other than RP, EBRT, PSI, or AS alone were excluded from this study.

Statistical Methods

All variables were categorical and are presented as frequencies (percentages). Chi-square tests were conducted to analyze the association between demographic variables and treatments. The graphs of practice pattern were generated by using the absolute counts and proportions of each treatment per year from 2004 to 2015. Series of univariate logistic regressions were conducted with treatment as the outcome and demographic variables as predictors to examine the factors that might affect the possibility of patients receiving NLT. Variables with p values less than 0.30 from univariate analysis were selected for multivariate logistic regression modal. The data were then further divided into 2 subsets: 2004 to 2009 and 2010 to 2015, to evaluate the potential change in predictors between the 2 periods. The same analyses were conducted for both data sets separately.

SAS Enterprise Guide 7.15 HF3 (SAS Institute, Inc) and SAS 9.4 were used for statistical analyses. All statistical analyses were performed at α = 0.05.

Results

Patient Characteristics

A total of 195,452 patients with LRPC diagnosed between 2004 and 2015 were identified in the NCDB, with age groups ranging from younger than 60 (freq (%): 57,137 [29%] to 75+ (20,116 [10%]) years, in 5-year increments. The racial distribution of the sample was African American (24,656,13%) and White (167,217,86%), with the remaining 2% as the other. Metropolitan centers treated 82% of patients (n = 160,092); similarly, most patients were treated in comprehensive community cancer programs (79,192, or 41%). Ninety-seven percent of the patients were insured through either private or government insurance (99,523 [51%] and 90,172 [46%], respectively). Among patients with LRPC, 45,462 (23.26%) patients received NLT, 58,501 (29.93%) received radical prostatectomy (RP), 44,833 (22.94%) received EBRT, and 46,654 (23.87%) received PSI (Table 1).

Table 1. - Summary of Patient Characteristics of Overall Study Population

Variable	n = 195,452
Age
<60	57,139 (29%)
60-64	42,140 (22%)
65-69	44,968 (23%)
70-74	31,095 (16%)
≥75	20,116 (10%)
Race
White	191,265 (85%)
AA	28,719 (13%)
Other	4039 (1.8%)
Area
Metropolitan	160,092 (82%)
Urban	26,937 (14%)
Rural	3477 (1.8%)
Facility type
Community Cancer Program	15,098 (7.8%)
Comprehensive Community Cancer Program	79,192 (41%)
Academic/Research Program	74,192 (38%)
Integrated Network Cancer Program	26,042 (13%)
Insurance status
Not insured	2611 (1.3%)
Private	99,523 (51%)
Government	90,172 (46%)
Income a
<30,000	28,783 (15%)
30,000-34,999	41,441 (21%)
35,000-45,999	51,375 (26%)
46,000+	72,932 (37%)
Education, % b
≥21	28,783 (14%)
13-20.9	45,310 (23%)
7-12.9	65,006 (33%)
<7	57,719 (30%)
Treatment
NLT	45,462 (23%)
RP	58,501 (30%)
EBRT	44,833 (23%)
PSI	46,656 (24%)
Mortality	24,545 (13%)

AA, African American; EBRT, external beam radiotherapy; NLT, no local therapy; PSI, prostate seed implant; RP, radical prostatectomy.

^aLevel of income by zipcode.

^bPercentage of high school graduates by zipcode.

Most patients with LRPC (79%) were managed in academic or comprehensive community cancer programs. The largest proportion of patients resided in metropolitan or urban areas. Most patients were covered by private or government insurance. Uninsured patients made up only 1.3% of the study population.

Among patients who received local therapies, 99.6% of those who received RP, 99.1% EBRT, and 99.9% PSI were treated within the first year of diagnosis, indicating that the corresponding local therapies were used as the primary mode of management rather than salvage treatment.

Treatment Modality Utilization

The distribution of age group, facility type, facility area, insurance status, education, race, and income differed significantly between treatment modalities (P < .0001 for all), indicating that the choices of management patients received differed significantly under each demographic variable. The all-cause mortality rates were 10%, 9.1%, 17%, and 15% for the NLT, RP, EBRT, and PSI groups, respectively (p < .0001) (Table 2).

Table 2. - Distribution of the Treatment Modalities Based on Patient Characteristics

Variable	NLT	RP	EBRT	PSI	P value
	(n = 45,462)	(n = 58,501)	(n = 44,833)	(n = 46,656)
Age
<60	12,190 (27%)	24,863 (43%)	8029 (18%)	12,055 (26%)	<.0001
60-64	9753 (21%)	13,489 (23%)	8511 (19%)	10,385 (22%)	<.0001
65-69	10,784 (24%)	11,086 (19%)	11,526 (26%)	11,571 (24%)	<.0001
70-74	7292 (16%)	5260 (9%)	10,100 (23%)	8442 (18%)	<.0001
>75	5443 (12%)	3803 (6.5%)	6667 (15%)	4203 (9%)	<.0001
Race
White	38,441 (85%)	52,367 (90%)	36,747 (82%)	39,658 (85%)	<.0001
AA	6055 (13%)	5186 (8.9%)	7171 (16%)	6242 (13%)	<.0001
Other	966 (2.1%)	948 (1.6%)	914 (2.0%)	756 (1.6%)	<.0001
Facility type
Community Cancer Program	3658 (8.1%)	3766 (6.5%)	4067 (9.1%)	3607 (7.7%)	<.0001
Comprehensive Community Cancer Program	13,161 (29%)	23,153 (40%)	20,086 (25%)	22,787 (49%)	<.0001
Academic/Research Program	23,749 (52%)	23,354 (40%)	15,086 (34%)	12,770 (27%)	<.0001
Integrated Network Cancer Program	4859 (11%)	8115 (14%)	5589 (12%)	7479 (16%)	<.0001
Area
Metro	37,416 (82%)	47,899 (82%)	37,404 (83%)	37,369 (80%)	<.0001
Urban	6085 (13%)	7805 (13%)	5778 (13%)	7268 (16%)	<.0001
Rural	793 (1.7%)	1129 (1.9%)	615 (1.4%)	940 (2%)	<.0001
Unknown	1168 (2.6%)	1668 (2.9%)	1036 (2.3%)	1079 (2.3%)	<.0001
Insurance status
Not insured	993 (2.2%)	617 (1.1%)	620 (1.4%)	381 (0.8%)	<.0001
Private	21,990 (48%)	37,306 (64%)	17,238 (38%)	22,985 (49%)	<.0001
Government	21,504 (47%)	19,779 (33%_	26,180 (58%)	22,707 (49%)	<.0001
Unknown	975 (2.1%)	799 (1.4%)	795 (1.8%)	583 (1.3%)	<.0001
Income a
<30,000	6549 (14%)	7396 (13%)	7627 (17%)	7207 (16%)	<.0001
30,000-34,999	9140 (20%)	12,090 (21%)	9561 (21%)	10,650 (23%)	<.0001
35,000-45,999	11,671 (26%)	15,907 (27%)	11,641 (26%)	12,154 (26%)	<.0001
46,000+	17,890 (40%)	22,875 (39%)	15,809 (35%)	16,358 (35%)	<.0001
a Education, %
≥21	6179 (14%)	6750 (12%)	7241 (16%)	6446 (14%)	<.0001
13-20.9	9620 (21%)	12,947 (22%)	11,127 (25%)	11,616 (25%)	<.0001
7-12.9	14,696 (32%)	19,830 (34%)	14,640 (33%)	15,838 (34%)	<.0001
<7	14,798 (33%)	18,770 (32%)	11,650 (26%)	12,500 (27%)	<.0001
Death	4762 (10%)	5351 (9.1%)	7485 (17%)	6947 (15%)	<.0001

AA, African American; EBRT, external beam radiotherapy; NLT, no local therapy; PSI, prostate seed implant; RP, radical prostatectomy.

^aDerived from the American Community Survey at the Zip code level.

Among the various treatment modalities, the utilization of NLT has steadily increased over time in absolute number and proportion (Figure 2), from 11.3% in 2004 to 53.5% in 2015. The use of RP peaked at 41.6% in 2008 but dropped to 17.6% in 2015. EBRT utilization peaked in 2006 (24.3%) and PSI in 2004 (35.3%), before falling to 18.1% and 10.8%, respectively, in 2015. During the same period, the number of LRPC cases peaked in 2008 at 21,089, before dropping steadily to 11,652 by 2014.

Figure 2.: Pattern of practice trends from 2004 to 2015. A, Absolute count of each treatment modality per year of diagnosis. B, Proportion of each treatment modality per year of diagnosis. EBRT, external beam radiotherapy; NLT, no local therapy; PSI, prostate seed implant; RP, radical prostatectomy.

Predictors of NLT Utilization

Univariate and multivariate logistic regression analyses were performed to examine the factors that could affect the decision to undergo NLT (Table 3). Age, ethnicity, insurance, income, education, and facility type were statistically significant predictors for patients receiving NLT in the univariate condition. As age increased, the tendency for NLT was higher, as represented by the odds ratio (OR). However, the OR was much higher for those older than 75 years, likely reflecting the strategy of WW, or observation for patients in this age bracket (OR 1.37, P < .0001). African American (AA) and other ethnicity groups were more likely to receive NLT (OR = 1.09 P < .0001, OR = 1.24, P <. 0001, respectively). Those who were uninsured and had an unknown insurance status were more likely to receive NLT than those who had private insurance with OR 2.16 (P < .0001) and 1.10 (P < .0001), respectively. Those with government insurance were more likely to receive NLT than those with private insurance, with OR 1.10 (P < .0001). Patients with lower income brackets, below $46,000, were less likely to receive NLT. Patients from regions where less than 7% residents have a high school diploma were more likely to receive NLT. Compared with the other programs, academic/research programs were more likely to offer NLT. For area of living, patients from urban areas were slightly less likely to receive NLT.

Table 3. - Univariate and Multivariate Logistic Regression Analyses on Association of Demographic Variables With NLT

Variable	Univariate		Multivariate
	OR (95% CI)	P value	OR (95% CI)	P value
Age
<60	Reference	—	—	—
60-64	1.11 (1.08, 1.14)	<0.0001	1.15 (1.11, 1.18)	<0.0001
65-69	1.16 (1.13, 1.20)	<0.0001	1.25 (1.20, 1.29)	<0.0001
70-74	1.13 (1.09, 1.17)	<0.0001	1.24 (1.20, 1.29)	<0.0001
>75	1.37 (1.32, 1.42)	<0.0001	1.56 (1.49, 1.63)	<0.0001
Race
White	Reference	—	—	—
AA	1.09 (1.06, 1.13)	<0.0001	1.10 (1.07, 1.14)	<0.0001
Other	1.24 (1.15, 1.33)	<0.0001	1.13 (1.04, 1.22)	0.0023
Insurance
Private	Reference	—	—	—
Not insured	2.16 (2.00, 2.35)	<0.0001	2.03 (1.87, 2.21)	<0.0001
Government	1.10 (1.08, 1.13)	<0.0001	1.02 (0.99, 1.06)	0.1210
Unknown	1.58 (1.46, 1.71)	<0.0001	1.36 (1.26, 1.47)	<0.0001
Income
$46,000+	Reference	—	—	—
<$30,000	0.91 (0.88, 0.94)	<0.0001	1.04 (0.99, 1.09)	0.0863
$30,000-$34999	0.87 (0.85, 0.90)	<0.0001	1.05 (1.01, 1.09)	0.0064
$35,000-$45999	0.90 (0.88, 0.93)	<0.0001	1.04 (1.01, 1.07)	0.0121
Education, %
≥21	Reference	—	—	—
13-20.9	0.89 (0.86, 0.92)	<0.0001	0.93 (0.89, 0.96)	0.0001
7-12.9	0.97 (0.93, 1.00)	0.0466	1.04 (1.00, 1.08)	0.0716
<7	1.14 (1.10, 1.18)	<0.0001	1.20 (1.14, 1.25)	<0.0001
Facility type
A/RP	Reference	—	—	—
CCP	0.69 (0.66, 0.72)	<0.0001	0.67 (0.64, 0.69)	<0.0001
CCCP	0.43 (0.42, 0.44)	<0.0001	0.42 (0.41, 0.43)	<0.0001
INCP	0.50 (0.48, 0.51)	<0.0001	0.49 (0.47, 0.51)	<0.0001
Area
Metro	Reference	—	—	—
Urban	0.96 (0.93, 0.99)	0.0053	1.08 (1.05, 1.12)	<0.0001
Rural	0.97 (0.89, 1.05)	0.4383	1.12 (1.03, 1.22)	0.0058
Unknown	1.01 (0.95, 1.08)	0.7181	1.10 (1.02, 1.18)	0.0120

AA, African American; ARP, Academic/Research Program; CCCP, Comprehensive Community Cancer Program; CCP, Community Cancer Program; CI, confidence limits; INCP, Integrated Network Cancer Program; OR, odds ratio.

Based on the results of univariate analysis, all demographic variables were selected to build a multivariate logistic regression model (Table 3). Similarly, NLT was more likely offered to patients who were older than 60 years and significantly more so to those who were older than 75 years. AA or other races were more likely to receive NLT compared with White. Patients of uninsured or unknown insurance status were more likely to receive NLT. Contrary to the univariate analysis, those with government insurance were no different from private insurance. Patients with incomes between $30,000 and $45,999, patients from areas with a lower level of education (<7%), patients from rural or unidentified area, and patients treated in settings other than academic/research programs were more likely to receive NLT.

Recognizing that the data set included patients across 2 distinct eras as NLT has become an increasingly popular choice since 2010 (Figure 2), we decided to divide the data set into 2 segments based on time, from 2004 to 2009 (Period 1) and from 2010 to 2015 (Period 2). We then performed the multivariate analyses for each period separately (Table 4). Comparing the results of multivariate analysis between the 2 periods, age remained a predictive factor with patients in age brackets older than 60 years receiving more NLT. However, there was a significant reduction in OR between the 2 periods, from 1.36 to 1.23 for the age bracket of 70 to 74 years and from 2.46 to 1.27 for the age bracket of older than 75 years. This is likely due to more younger patients receiving NLT (reference younger than 60 years) from 2010 through 2015. More interestingly, AA patients were 17% more likely than White patients to receive NLT during the period of 2004 to 2009 (OR 1.17, P < .0001) but became 6% less likely than White patients to receive NLT during the period of 2010 to 2015 (OR 0.94, P = .0039), likely reflecting increasing utilization of NLT among White patients and possibly in AA patients as well. Patients of other races were more likely to receive NLT than White during the period of 2010 to 2015 (OR 1.14, P = .0142). Insurance status remained a predictive factor with uninsured or unknown insurance status leading to a higher chance of NLT. However, the OR reduced between the 2 periods from 2.02 to 1.89 for the uninsured and from 1.85 to 1.30 for the unknown insurance status, respectively, likely reflecting that more patients with private insurance (reference) received NLT between 2010 and 2015. Patients with incomes less than $46,000+ (reference) were more likely to receive NLT during the period of 2010 to 2015. Patients from areas with lower education levels (<7% and 7%-12.9%) received more NLT. Patients from urban, rural, and unidentified areas were more likely to receive NLT compared with metropolitan areas. Compared with academic/research program, the rest of the programs continued to have lower tendency to offer NLT and the OR was further reduced between the 2 periods, from 0.82 to 0.55 for community cancer program (CCP), from 0.51 to 0.38 for comprehensive community cancer program (CCCP), and from 0.61 to 0.44 for integrated network cancer program, respectively. This is most likely due to the fact that academic/research programs (A/RPs) were the increasingly dominant source of offering NLT and leading the practice change between 2010 and 2015.

Table 4. - Univariate and Multivariate Logistic Regression Analyses on Association of Demographic Variables With NLT in 2 Different Periods

Variable	2004-2009				2010-2015
	Univariate		Multivariate		Univariate		Multivariate
	OR (95% CI)	P value	OR (95% CI)	P value	OR (95% CI)	P value	OR (95% CI)	P value
Age
<60	Reference	—	—	—	Reference	—	—	—
60-64	1.07 (1.01, 1.13)	0.0195	1.09 (1.04, 1.16)	0.0014	1.08 (1.03, 1.12)	0.0003	1.10 (1.06, 1.15)	<0.0001
65-69	1.14 (1.08, 1.20)	<0.0001	1.19 (1.12, 1.27)	<0.0001	1.11 (1.07, 1.16)	<0.0001	1.22 (1.17, 1.28)	<0.0001
70-74	1.28 (1.21, 1.35)	<0.0001	1.36 (1.27, 1.46)	<0.0001	1.08 (1.03, 1.13)	0.0010	1.23 (1.17, 1.30)	<0.0001
>75	2.25 (2.13, 2.38)	<0.0001	2.46 (2.30, 2.64)	<0.0001	1.06 (1.01, 1.12)	0.0323	1.27 (1.19, 1.35)	<0.0001
Race
White	Reference	—	—	—	Reference	—	—	—
AA	1.16 (1.10, 1.23)	<0.0001	1.17 (1.11, 1.24)	<0.0001	0.90 (0.87, 0.94)	<0.0001	0.94 (0.90, 0.98)	0.0039
Other	1.09 (0.96, 1.25)	0.1862	0.98 (0.86, 1.13)	0.7991	1.22 (1.11, 1.35)	<0.0001	1.14 (1.03, 1.26)	0.0142
Insurance
Private	Reference	—	—	—	Reference	—	—	—
Not insured	2.20 (1.92, 2.53)	<0.0001	2.02 (1.75, 2.33)	<0.0001	1.90 (1.71, 2.12)	<0.0001	1.89 (1.69, 2.11)	<0.0001
Government	1.32 (1.27, 1.37)	<0.0001	1.05 (1.00, 1.11)	0.0622	0.97 (0.95, 1.00)	0.0714	0.95 (0.91, 0.99)	0.0099
Unknown	2.28 (2.04, 2.54)	<0.0001	1.85 (1.65, 2.08)	<0.0001	1.40 (1.25, 1.57)	<0.0001	1.30 (1.15, 1.46)	<0.0001
Income
$46,000+	Reference	—	—	—	Reference	—	—	—
<$30,000	1.04 (0.99, 1.10)	0.1513	1.04 (0.97, 1.12)	0.3145	0.77 (0.74, 0.81)	<0.0001	1.07 (1.00, 1.13)	0.0408
$30,000-$34999	0.93 (0.88, 0.98)	0.0027	1.01 (0.95, 1.07)	0.7181	0.81 (0.78, 0.84)	<0.0001	1.11 (1.05, 1.16)	<0.0001
$35,000-$45999	0.97 (0.93, 1.02)	0.2263	1.04 (0.99, 1.09)	0.1558	0.85 (0.82, 0.88)	<0.0001	1.06 (1.02, 1.10)	0.0063
Education, %
≥21	Reference	—	—	—	Reference	—	—	—
13-20.9	0.80 (0.76, 0.86)	<0.0001	0.83 (0.78, 0.89)	<0.0001	0.99 (0.94, 1.04)	0.5834	1.02 (0.97, 1.07)	0.5447
7-12.9	0.87 (0.82, 0.92)	<0.0001	0.92 (0.86, 0.99)	0.0231	1.12 (1.07, 1.17)	<0.0001	1.18 (1.12, 1.25)	<0.0001
<7	0.95 (0.90, 1.01)	0.1023	1.01 (0.94, 1.09)	0.7577	1.39 (1.33, 1.45)	<0.0001	1.40 (1.32, 1.49)	<0.0001
Facility type
A/RP	Reference	—	—	—	Reference	—	—	—
CCP	0.90 (0.84, 0.96)	0.0010	0.82 (0.77, 0.88)	<0.0001	0.55 (0.52, 0.58)	<0.0001	0.55 (0.52, 0.58)	<0.0001
CCCP	0.54 (0.52, 0.56)	<0.0001	0.51 (0.49, 0.53)	<0.0001	0.39 (0.37, 0.40)	<0.0001	0.38 (0.37, 0.39)	<0.0001
INCP	0.62 (0.58, 0.65)	<0.0001	0.61 (0.57, 0.64)	<0.0001	0.44 (0.42, 0.46)	<0.0001	0.44 (0.42, 0.46)	<0.0001
Area
Metro	Reference	—	Not included		Reference	—	—	—
Urban	0.95 (0.90, 1.00)	0.0721			0.91 (0.88, 0.95)	<0.0001	1.11 (1.06, 1.16)	<0.0001
Rural	0.93 (0.81, 1.07)	0.3330			0.94 (0.84, 1.04)	0.2081	1.21 (1.08, 1.35)	0.0006
Unknown	1.03 (0.93, 1.15)	0.5693			1.07 (0.98, 1.18)	0.1264	1.13 (1.03, 1.24)	0.0138

AA, African American; ARP, Academic/Research Program; CCCP, Comprehensive Community Cancer Program; CCP, Community Cancer Program; CI, confidence limits; INCP, Integrated Network Cancer Program; OR, odds ratio.

Discussion

We evaluated the treatment trends for various treatment modalities using the NCDB data. This is the largest study to date that has focused on LRPC, involving nearly 200,000 patients. Between 2004 and 2015, there was a significant shift in practice patterns. As a trendsetter, AS (as represented by NLT in this study) has become the most common management strategy for LRPC.

AS, also known as expectant management or deferred treatment, involves closely monitoring patients in anticipating of administering curative local treatment if the disease progresses, to minimize “overtreatment” and treatment-associated adverse effects. It typically involves younger patients and requires scheduled biopsies and imaging studies. WW was used interchangeably in the past, but some in the field nowadays believe that it more implies observation with PSA testing no more than twice a year, an approach typically used for older patients. In this study, most patients were of relatively younger ages, with 74% younger than 70 years and 90% younger than 75 years (Table 1). Although it would be difficult to distinguish AS from WW or no treatment for other reasons in this study, given that most patients were relatively young with no or minimal comorbidity, it is safe to conclude that NLT likely reflects the AS approach, which has become the dominant mode of LRPC management by 2015. This trend started in 2009 and has accelerated since then (Fig 2). This is most likely a consequence of several influential studies on AS and subsequent societal recommendations.^5,6,15-18

Based on our multivariate analysis, as led by academic/research programs, AS had increasingly been offered to younger patients including those younger than 60 years. Race was a factor for the period of 2004 to 2009 with more AA offered for NLT, but racial differences for NLT utilization became less apparent with the more widespread adoption of AS across racial groups during the period of 2010 to 2015. Yet, besides the “good reasons” for AS, the multivariate analysis also shed light on other reasons for offering NLT, mainly socioeconomic status. Patients with lower socioeconomic status such as those of lower income brackets, living in areas of fewer high school graduates, being uninsured or having unknown insurance status, or living in rural or urban inner cities were less likely to be offered local treatments but NLT. Furthermore, except for the A/RP, the rest of the facilities, including CCCP, CCP, and INCP, were more inclined to offer local treatments than NLT, even during the period of 2010 to 2015, a phenomenon that may partly be driven by practice economics but also possibly the “trickling down” impact of societal guidelines that may take years to affect community practice. In a study published in 2017, Parikh et al¹⁹ focused on 40,839 patients with very low-risk prostate cancer from 2010 to 2013 and noted similar socioclinical disparity and overutilization of local therapies as opposed to NLT in the very low-risk prostate cancer population.

In a recent edition of the NCCN guidelines in late 2021, the “preferred” status was once removed for AS,²⁰ which was later reinstated,¹³ but rekindled debate and interest in the management of LRPC. The current AUA/ASTRO/SUO guidelines recognize AS as a safe management option for LRPC, particularly in patients with very low-risk prostate cancer.¹⁴ Although it is unlikely to improve oncologic outcomes on its own, it preserves quality of life by avoiding treatment-related side effects, which is especially important given the widespread belief that there is no single best treatment modality for LRPC. Several studies on the AS approach have demonstrated that it is safe and feasible over a period of 5 to 10 years,^5,7-9,15 especially for a fraction of LRPC, those with very low-risk prostate cancer, as defined by Epstein et al.¹⁰ However, the long-term natural history of AS for the rest of the LRPC remains unclear. Discordant pathology between prostate biopsy and prostatectomy is a known phenomenon. A recent study suggested that there is an increasing rate of pathologic upgrading in LRPC in the era of AS.²¹ Klotz et al²² reported that the 10-year and 15-year actual cause-specific survival rates in a single arm prospective AS cohort were 98.1% and 94.3%, respectively. It is worthwhile to mention that more than 99% of patients in this study received their local therapies within the first year of diagnosis, indicating that the treatment of choice as the primary rather than a deferred mode of management.

An earlier observation in 2015 by Weiner et al²³ noted the beginning of the trend of increasing utilization of AS from 13% to 20% by using NCDB and from 21% to 32% by using the Surveillance, Epidemiology and End Results (SEER) data among LRPC diagnosed between 2004 and 2010. Our study demonstrated that there has been a more significant shift in the practice pattern in LRPC. AS is currently the most popular choice of management for LRPC. In 2015, approximately 53.5% of patients with LRPC were managed with AS, compared with approximately 17.6% with RP. This is the polar opposite of the practice pattern observed in 2004. It remains to be seen whether such a dramatic shift in trend will have an effect on the overall survival of the LRPC population. This is most emphatically outside the scope of this study. Hamdy et al¹¹ showed no difference in prostate-specific mortality among patients managed with AS, RP, or EBRT over a median follow-up of 10 years in the ProtecT trial population among which 77% of patients had tumor of Gleason score of 6.

The strength of this study is its large sample size, which is the largest to date, resulting in greater statistical power. By including only those with low Charlson-Deyo comorbidity score, the distribution of each treatment group is even and comparable. However, owing to the nature of the NCDB study, there were some limitations. This study is retrospective in nature. Patients in each cohort were not randomized, so selection bias likely influenced the distribution of patients among different treatment options, despite the fact that we only included patients with a Charlson-Deyo comorbidity score of 0 or 1, implying that comorbidity had a lower influence on treatment decisions. Furthermore, we were unable to distinguish AS from WW, although NLT for younger patients in academic institutions most likely represents AS. For those older than 75 years, WW might be more likely than AS. However, for those between 70 and 74 years, it is would be more challenging to determine the intention of NLT, be it WW vs AS, due to the limited information provided in the NCDB data set. The Gleason scoring methods were modified at the 2005 consensus conference of international experts in urological pathology. The changes resulted in some previously thought to be low Gleason scores being upgraded to a higher score, 7 or higher, resulting in risk group migration.^24,25 Our patient population spans from 2004 to 2015. As a result, it is possible that some patients diagnosed before 2005 may have been classified as Gleason 7 under the modified Gleason system. It would be difficult to determine the impact of these changes on the patient population and conclusion of this study. Finally, we cannot assess the impact on survival of LRPC because of current practice trend changes favoring AS, and it is likely too early to do so.

Conclusion

There has been a significant paradigm shift in LRPC treatment, signified by the trend in which AS has risen to become the preferred choice of management. This trend is mostly led by academic institutions, and NLT was more widely used among younger patients across racial groups, reflecting the sea change in the management philosophy of LRPC across the country. However, socioeconomic status and practice economics continued to play a role in selecting treatment modalities for LRPC.

Funding

No funding was received for this study.

Conflicts of Interest

S.M. received honorarium payment from Sanofi, AstraZeneca, Pfizer, Seagen, Astellas, Bayer, Exelisix, and Cardinal Health for consultancy and from Bristol-Myers Squibb for the speaker bureau in the past. But none of the consultancy and speaker bureau activities are relevant to the content of this manuscript. The rest of the authors have no conflicts of interest to declare.

Author Contributions

S.M.: conceptualization, methodology, formal analysis, writing original draft, critical review and editing, supervising. A.S.: conceptualization, methodology, formal analysis, writing original draft, critical review and editing. Y.Y.: methodology, statistical analysis, writing, critical review and editing. R.S.: methodology, statistical analysis, writing, critical review and editing. R.E.W.: methodology, formal analysis, writing, critical review and editing. R.F.: writing, critical review and editing. R.M.: writing, critical review and editing. J.L.: writing, critical review and editing. A.S.: writing, critical review and editing. J.C.: conceptualization, writing, critical review and editing, supervising.

Declaration

There was no funding for this study. This study has not been published in part or full from elsewhere.

Ethical Statement

A retrospective review was performed using deidentified patient data. Given patient deidentification, this study was exempt from institutional review board oversight.