World Journal of Urology

, Volume 32, Issue 4, pp 959–964

Application of the 2013 American Urological Association early detection of prostate cancer guideline: Who will we miss?

Authors

  • Gregory B. Auffenberg
    • Department of UrologyNorthwestern University Feinberg School of Medicine
    • Department of UrologyNorthwestern University Feinberg School of Medicine
Original Article

DOI: 10.1007/s00345-014-1341-2

Cite this article as:
Auffenberg, G.B. & Meeks, J.J. World J Urol (2014) 32: 959. doi:10.1007/s00345-014-1341-2
  • 200 Views

Abstract

Purpose

The American Urological Association (AUA) published new prostate cancer (CaP) screening guidelines in 2013. We apply the guidelines to a retrospective cohort to compare tumor characteristics of those no longer recommended for screening with those who remain screening candidates.

Methods

We identified cases of screening detected CaP (stage cT1c) in the Surveillance Epidemiology and End Results database from October 2005 to December 2010. The 2013 AUA Guidelines were retrospectively applied to the cohort. Men were categorized into three groups for comparison based on whether or not they would now be recommended for CaP screening (Unscreened, Young Unscreened, and Screened). We compared clinical and pathological characteristics of CaP across study groups.

Results

A total of 142,382 men were identified. Screening would no longer be recommended for 40,160. Those no longer recommended for screening had higher median PSA (6.4 vs. 5.8 ng/mL, p < 0.01), more Gleason 7 and ≥8 CaP on prostate biopsy (36.4 vs. 34.8 %, p < 0.001; 12.4 vs. 9.2 %, p < 0.001, respectively) and slightly more Gleason ≥8 CaP (9.0 vs. 7.5 %, p = 0.03), and T3 tumors (17.3 vs. 16.5 %, p = 0.01) at prostatectomy. Nodal and distant metastasis rates were clinically equivalent among men screened and unscreened. Subgroup analysis of young patients (40–54 years old) no longer recommended for screening identified intermediate or high-risk Gleason scores at prostatectomy 57.6 % of the time.

Conclusions

Features of CaP in men no longer recommended for routine screening are largely equivalent to if not worse than those in screened men.

Keywords

Prostate cancerScreeningGuidelinesProstate-specific antigen

Introduction

Prostate cancer (CaP) screening with digital rectal examination (DRE) and serum prostate-specific antigen (PSA) testing has been widely adopted in the USA to diagnose early-stage CaP [1]. Despite declining mortality rates from CaP in the PSA-era, the United States Preventive Service Task Force (USPSTF) recommended against routine screening for CaP in 2012 [2]. This recommendation was based on evidence from five international randomized controlled trials (RCTs) investigating PSA-based screening [37]. The USPSTF stated “the benefits of PSA-based screening for prostate cancer, as currently used and studied in randomized, controlled trials do not outweigh the harms” [2]. The initial response to the USPSTF statements from the urological community, including the American Urological Association (AUA) and American Association of Clinical Urologists, was in opposition to the USPSTF recommendations [810].

Amid the ongoing debate about PSA-based CaP screening, the AUA recently published the 2013 AUA Early Detection of Prostate Cancer Guideline [11]. An expert panel authored the 2013 AUA guideline after independent consideration of the five prospective RCTs evaluated by USPSTF. The 2013 AUA guideline recommends against PSA screening in all men ≤40 and does not recommend routine screening of men at average risk between ages 40 and 54. Screening of men at higher risk in the 40–55-year-old range, specifically Black men and those with a family history of CaP, was not described beyond the statement that their “screening should be individualized”. The 2013 AUA guideline does recommend screening for men 55–69 years old if desired by a patient after a shared decision-making process is employed, but does not recommend screening in men age ≥70 or any man with <10–15 years life expectancy [11].

The 2013 AUA guideline’s-specific recommendation against screening in two age groups, those <55 and >69, is different from the organization’s prior guidance. The 2009 AUA PSA Best Practice Statement [12] called for a baseline PSA at age 40 and for future screening intervals to be determined based on that result. The decision to no longer recommend screening for average risk men <55 in the 2013 Guideline was explained to be largely due to the lack of guidance from the RCTs for men in this age group [13]. For men 70 or older, in 2009, the AUA stated that “the benefits of screening may decline rapidly with age” but did not provide a specific age at which screening should be stopped. In 2013, the AUA panel felt the evidence no longer supported screening these older men.

Given the USPSTF statements and the 2013 AUA Guidelines, it is likely that the number of American men screened for CaP will decrease. In order to preliminarily evaluate the impact of the 2013 AUA guideline, we retrospectively model the application of the guidelines in 40- to 75-year-old men with screening detected CaP diagnosed between 2005 and 2010 in a large American cohort. We compare the tumor characteristics of patients who would still be screening candidates under the 2013 guidelines with those who are no longer recommended for screening.

Patients and methods

Database

Using the Surveillance Epidemiology and End Results (SEERs) database November 2012 release [14], we identified patients and retrospectively gathered cancer-related information for analysis.

Cohort

Included in the study population were men 40–75 years old diagnosed with adenocarcinoma of the prostate between October 2005 and December 2010. Clinical stage was limited to T1c (SEER category CS Extension–Clinical Extension = T1c) in order to ensure all malignancies were discovered as a result of CaP screening. The dataset was limited to include only cases diagnosed after October 2005, the date of publication of the widely accepted updated Gleason grading system, in order to maintain homogeneity of the pathologic data [15].

Study groups

The cohort of men with prostate cancer was divided into three study groups for comparison: Unscreened, Young Unscreened, and Screened (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00345-014-1341-2/MediaObjects/345_2014_1341_Fig1_HTML.gif
Fig. 1

Makeup of study groups

The Unscreened group represents patients who were candidates for CaP screening based on the 2009 AUA best practice statements but are no longer recommended for routine screening in the 2013 Guideline [11, 12]. The Unscreened group contains men age 40–54 at average risk and all men 70–75 years old. Black men and those with a family history of CaP were excluded from this group, because they are specifically referenced as men at greater than average risk in the 40–54-year-old age range. SEER does not contain data on family history, so we used modeling to account for family history in non-black men 40–54 years old. To model, we assumed the rate of CaP family history in SEER would be similar to the higher estimates in the literature (30 %) [16, 17]. Thus, a randomly selected 30 % of non-black men 40–54 years old were excluded from the Unscreened group and included in the Screened group.

The Young Unscreened group is a subset of patients from the Unscreened cohort. It contains the men from the Unscreened cohort with age restricted to 40–54 years.

The Screened cohort represents men who are recommended for routine screening in the 2013 AUA Guideline. Thus, this cohort included all men age 55–69 and all men with greater than average risk aged 40–54 years (i.e., Blacks and those with modeled “family history”).

Analysis

The clinical and pathological staging elements analyzed were based on prostate cancer-specific factors entered into the SEER database (“Appendix” contains data categories analyzed). Gleason scores were used to categorize pathologic elements as low risk (score ≤ 6), intermediate risk (score = 7), or high risk (score ≥ 8).

Statistical testing was performed using SAS (SAS institute, Cary, NC, USA) and Excel (Microsoft Corporation, Redmond, WA, USA). Chi-squared and Wilcoxon rank-sum test comparative statistics were used where indicated. Within the cohort, data were not available for every patient for every element analyzed. In instances where patients had missing data, they were excluded from the analysis of that specific element.

Results

A total of 142,382 men meeting the study criteria were identified, 102,222 (71.8 % of total cohort) patients were included in the Screened group, 40,160 (28.2 % of total cohort) in the Unscreened group, and 10,052 in the Young Unscreened group (25.0 % of Unscreened patients 10,052/40,160). Demographic data are in Table 1.
Table 1

Demographic and clinical data

 

Young Unscreened

All Unscreened

Screened

n

Group (%)

p value

n

Group (%)

p value

n

Group (%)

Total

10,052

  

40,160

  

102,222

 

Race

 American Indian/Alaskan

42

0.4

 

122

0.3

 

327

0.3

 Asian/Pacific Islander

392

3.9

 

2,031

5.1

 

4,058

4.0

 Black

0

0.0

 

3,899

9.7

 

19,666

19.2

 Other

54

0.5

 

112

0.3

 

288

0.3

 White

9,260

92.1

 

33,011

82.2

 

75,913

74.3

 Unknown

304

3.0

 

985

2.5

 

1,970

1.9

Age (years)

 Mean

51

 

<0.01

70

 

<0.01

62

 

 Median

51

 

<0.01

71

 

<0.01

62

 

PSA

9,272

  

37,089

  

95,183

 

 Mean (ng/mL)

7.8

 

<0.01

10.1

 

<0.01

9.1

 

 Median (ng/mL)

5.2

 

<0.01

6.4

 

<0.01

5.8

 

Gleason score of diagnostic biopsy

2,365

  

9,384

  

24,936

 

 ≤6

1,525

64.5

<0.001

4,804

51.2

<0.001

13,955

56.0

 7

714

30.2

<0.001

3,414

36.4

<0.001

8,686

34.8

 ≥8

126

5.3

<0.001

1,166

12.4

<0.001

2,295

9.2

Gleason score of surgical or autopsy specimen

1,583

  

2,852

  

11,183

 

 ≤6

672

42.4

<0.001

981

34.4

0.03

3,869

34.6

 7

831

52.5

<0.001

1,615

56.6

0.03

6,478

57.9

 ≥8

80

5.1

<0.001

256

9.0

0.03

836

7.5

Pathologic stage

6,852

  

12,058

  

47,623

 

 ≤T2

5,930

86.5

<0.001

9,897

82.1

0.01

39,398

82.7

 T3

889

13.0

<0.001

2,090

17.3

0.01

7,853

16.5

 T4

33

0.5

<0.001

71

0.6

0.01

372

0.8

Size of dominant tumor

2,107

  

4,214

  

15,190

 

 Mean (mm)

18.0

 

<0.001

19.9

 

<0.001

19.0

 

 Median (mm)

15.0

 

1

15.0

 

1

15.0

 

 SD

39.3

  

47.9

  

44.5

 

Regional node status

3,508

  

6,857

  

26,372

 

 Negative

3,433

97.9

0.02

6,659

97.1

0.68

25,635

97.2

 Positive

75

2.1

0.02

198

2.9

0.68

737

2.8

Metastatic status

9,916

  

39,321

  

100,659

 

 No distant mets

9,829

99.1

0.28

38,793

98.7

<0.001

99,613

99.0

 Distant/other mets

25

0.3

0.28

163

0.4

<0.001

328

0.3

 Bone

62

0.6

0.28

365

0.9

<0.001

718

0.7

PSA and tumor size compared using Wilcoxon rank-sum test, Gleason scores, tumor extension, nodal status and metastatic status compared using chi-squared analysis, p values represent comparison to the Screened group, p < 0.05 considered significant. Percentages calculated relative to the n for a given data category which is displayed into row of category

Unscreened men had a higher median PSA than Screened men (Table 1). Overall, 48.8 % (4,580/9,384) of Unscreened patients with available prostate biopsy pathology had intermediate or high-risk Gleason scores on biopsy. Gleason 7 CaP was found in 36.4 % (3,414/9,384) of Unscreened men and an additional 12.4 % (1,166/9,384) had high-risk Gleason scores (≥8) on biopsy. In Young Unscreened men, 35.5 % (840/2,365) of patients had intermediate or high-risk Gleason scores on prostate biopsy (Table 1).

Unscreened men that had surgical removal of their prostate had higher rates of pathologic T3 tumors and a higher frequency of Gleason ≥8 tumors compared to Screened (Table 1). Intermediate and high-risk Gleason scores were found in 57.6 % (911/1,583) of Young Unscreened patients at prostatectomy (Table 1). There was no difference in rates of nodal metastases across Unscreened and Screened men. Mean tumor size, and rates of bone and distant metastases were statistically different but clinically indistinguishable between the Screened and Unscreened cohorts (Table 1).

Discussion

Several investigators have shown a decrease in CaP screening following the USPSTF recommendations, and it is possible the AUA’s 2013 guidelines will support this trend [18, 19]. The 2013 screening guidelines were based on the published RCTs conducted on PSA screening [37]. Despite the application of the highest level evidence available, four of the AUA’s five guideline statements were assigned an evidence strength of C (low strength) and one was given evidence strength of B (moderate), suggesting a dearth of quality evidence for or against a real benefit to CaP screening. Less screening may reduce the number of patients with insignificant disease exposed to the potential harms of diagnosis and treatment. Alternatively, decreased screening may delay diagnoses and ultimately lead to an increase in late stage CaP. Due to the relatively slow moving nature of most CaP, the impact of the new guidelines will not be determined for decades.

Although we may not know the true effect of the 2013 guidelines for some time, our analysis of a large cohort of screening detected CaP cases suggests that 28.2 % (40,160/142,382) of these cases may not have been diagnosed as expeditiously under the 2013 guidelines. Much of the potentially undiagnosed cancer was significant CaP. For instance, across the Unscreened and Screened populations the prevalence of Gleason ≥7 tumors in surgical pathology specimens was statistically different but clinically near identical (65.6 vs. 65.4 %, respectively). Gleason score ≥7 on surgical pathology is associated with a higher risk of biochemical recurrence, metastasis, and cancer-specific mortality [2023]. Thus, in our cohort, the Unscreened group appears to be at similar pathologic risk of poor outcomes as those still recommended for screening.

Shifting to consider just the young men (<55 years old) in our cohort and the population at large, unfortunately there is little prospective RCT evidence to guide screening decisions in these men. The lack of level-one evidence regarding the management of these men has resulted in dichotomous recommendations for screening. In the absence of prospective data as mentioned, the AUA recommended that average risk younger men should not be screened. Alternatively, after considering the RCTs along with observational data the European Association of Urology recently recommended a baseline PSA at age 40 and a subsequent decision about screening interval based on the result [24]. Similarly, the authors of the Melbourne Consensus Statement call for baseline PSA testing for men in their 40s [25]. It deserves mention that overdiagnosis and overtreatment of younger men in the setting of indolent disease may expose men to the potential ill effects of screening and treatment for an extended time period given their age, and this underscores the importance of prudence when determining whether to screen or treat this group. However, there are several observational reports that conclude screening may be of value in this group. For instance, Loeb et al. [26] concluded that an elevated baseline PSA at a young age remains an important prognostic factor for identifying CaP mortality. In an analysis of patients from the Malmo Preventive Project, Vickers et al. [27] demonstrate that 44 % of CaP deaths occurred in the 10 % of men with the highest PSA levels at age 45–49. Although observational in nature, these studies represent the best available evidence we have when it comes to decisions about screening younger men.

In evaluating patients 40–54 years of age, our analysis demonstrated that these patients had lower median PSA at diagnosis, lower rates of intermediate and high-risk Gleason scores on biopsy and surgical specimens, and lower rates of T3/T4 tumors compared to the Screened patients. Although rates of adverse clinical parameters are less in the Young Unscreened, it remains that 57.6 % (911/1,583) of examined prostates had tumors with intermediate or high-risk Gleason scores and 13.5 % (922/6,852) had pT3 or pT4 tumors. If CaP screening is eliminated for this group, it is unknown whether the diagnosis of these tumors would be delayed until 55 years of age without ill effect or whether patients would have worse outcomes due to delayed diagnoses.

The potential harms of prostate cancer screening and treatment are an important consideration and had a significant impact on the USPSTF and AUA recommendations. As the AUA guideline notes, the rate of minor complications from biopsy is 20–50 %, 30-day risk of post-biopsy hospitalization is 4 %, and the estimated “overdiagnosis” rates of CaP vary from 5 to 75 % [11]. Notable psychological impacts of screening and treatment have been documented as well. [28, 29] Additionally, overdiagnosis may put patients at risk for overtreatment and at risk for common treatment side effects when the disease may have been of minimal impact. This should be closely considered by patient and physician when making screening and treatment decisions.

The SEER database contains information on a representative 28 % of the US population [14]. Extrapolating our data to model the number of cases missed across 100 % of the US population between October 2005 and December 2010 demonstrates that 143,429 T1c CaP patients would have been at risk for delayed detection of CaP under the new guidelines if universally applied. Of those men, 35,900 would be in the 40–54-year-old age range. That translates to 27,320 cases of T1c prostate cancer cases at risk for delayed diagnosis per year with 6,838 cases/year being in men <55 years old.

Our study is limited by the inherent biases of a retrospective design. Additionally, data were not available in every study category for all patients, variably reducing the cohort size in each category. The chosen inclusion dates of October 2005–December 2012 were used to develop the most contemporary cohort possible from SEER (a dataset containing cases dating back to 1973) while maintaining as much pathological homogeneity in reported results as possible (i.e., all specimens evaluated after date of updated Gleason Scoring system in October 2005). Although it is likely there was not universal adoption of the pathologic guidelines at the date of publication, there is no way to quantify rate of adoption, and thus, the date of publication was used to limit the cohort. SEER does not provide information on the number of patients screened per case diagnosed, so number needed to screen and treat calculations are not possible. The method for modeling family history via random selection may lead to an inaccurate estimation of the CaP burden across the study groups. However, it is worth noting that a large published cohort has shown no significant difference in clinicopathological features in patients with a family history of CaP compared to those without [30]. Thus, although the modeling method we use may slightly misjudge the cancer prevalence in our study groups, it is likely the pathological features remain similar across groups despite this. Additionally, to be included in our cohort, patients had to have clinical stage T1c CaP. We did this to ensure all patients were screening detected since T1c by definition only contains PSA screening detected cancer. Not all screening detected cases are clinical stage T1c. We likely underestimate the rate of the screening detected cancer in the database by limiting the cohort in this way, but this was the only way to ensure all cases are screening detected within the limitations of the dataset. Lastly, our conclusions are based on surrogate markers of disease severity (lab values, pathologic, and radiographic findings). Longer follow-up duration would be needed to meaningfully apply survival statistics to these patients diagnosed after October 2005. Despite these limitations, the clinical and pathologic data provide useful information regarding the CaP burden across study groups and the US population as a whole.

In conclusion, we estimate that applying the 2013 AUA guidelines nationwide from October 2005 through 2010 may have delayed the detection of 143,429 T1c CaP cases due to lack of screening (equal to 27,320 patients every year). Of potentially Unscreened men who underwent prostatectomy, 65.4 % had Gleason ≥7 at prostatectomy. In 40–54-year-old Unscreened CaP patients, 57.6 % had Gleason score ≥7 at prostatectomy, and 13.5 % were pT3 or pT4. Excluding these men with higher-grade disease from screening may ultimately result in delayed detection and increased incidence of incurable CaP.

Acknowledgments

Special thanks to Mr. John Cashy for assistance with statistical analysis and data acquisition.

Conflict of interest

The authors certify that there is no actual or potential conflict of interest in relation to this article.

Ethical standards

All patient data in this study were acquired through the fully de-identified SEER database, which is collected and disseminated via the National Cancer Institute in a manner compliant with the 1964 Declaration of Helsinki and later amendments.

Copyright information

© Springer-Verlag Berlin Heidelberg 2014