Adverse upgrading and/or upstaging in contemporary low-risk prostate cancer patients

Background Upgrading and/or upstaging in low-risk prostate cancer (PCa) patients may represent an indication for active treatment instead of active surveillance (AS). We addressed contemporary upgrading and/or upstaging rates in a large population based-cohort of low-risk PCa patients. Materials and methods Whitin the SEER database (2010–2015), NCCN low-risk PCa patients were identified across management modalities: radical prostatectomy (RP), radiotherapy (RT) and non-local treatment (NLT). In RP patients, upgrading and/or upstaging rates were assessed in logistic regression models. Results Overall, of 27,901 low-risk PCa patients, 38% underwent RP vs 28% RT vs 34% NLT. RP patients were the youngest and harbored the highest percentage of positive cores and a higher rate of cT2a than NLT. At RP, 46.2% were upgraded to GGG ≥ 2, 6.0% to GGG ≥ 3 and 10.5% harbored nonorgan-confined stage (NOC, pT3-4 or pN1). Of NOC patients, 1.6% harbored GGG ≥ 3, 6.3% harbored GGG2 and 2.6% harbored GGG1. Of pT2 patients, 4.4% harbored GGG ≥ 3, 33.9% harbored GGG2 and 51.3% harbored GGG1. Age, PSA, percentage of positive cores and number of positive cores independently predicted the presence of NOC and/or GGG ≥ 3, but with low accuracy (63.9%). Conclusions In low-risk PCa, critical changes between tumor grade and stage at biopsy vs RP may be expected in very few patients: NOC with GGG ≥ 3 in 1.6% and NOC with GGG2 in 6.3%. Other patients with upgrading and/or upstaging combinations will invariably harbor either pT2 or GGG1 that far less critically affect PCa prognosis.


Introduction
North American and European guidelines recommend active surveillance (AS) in low-risk prostate cancer (PCa) patients, based on comparable long-term oncologic outcomes relative to active treatment [1][2][3][4] . However, these guidelines are based on findings of historical studies 5-8 when conservative management rates (active surveillance/active monitoring/watchful waiting) were substantially lower than currently 9,10 . Moreover, grading of PCa might have not been the same, since in contemporary patients the combination of systemic and target biopsy has shown to significantly reduce upgrading rates [11][12][13][14] . Similarly, contemporary use of MRI prior to prostate biopsy decreased the rate of clinically insignificant PCa diagnoses at initial biopsy [15][16][17] . In consequences historical data may not be applicable to contemporary patients. To address this void, we tested most contemporary upgrading and/or upstaging rates with specific focus on critical upgrading and/or upstaging, respectively, defined as GGG ≥ 3 and/or nonorgan-confined stage. We hypothesized that contemporary upgrading and/or upstaging rates were substantially lower than historical rate.

Study population
The Surveillance, Epidemiology and End Results (SEER) database samples 26% of the USA and approximates the USA in terms of geographic and demographic composition, as well as cancer incidence 18 . Within the SEER database spanning years 2010-2015, we identified all non-metastatic patients, aged between 40 and 75 years, with histologically confirmed adenocarcinoma of the prostate, diagnosed at biopsy (International Classification of Disease for Oncology [ICD-O-3] code 8140 site code C61.9), who fulfilled the NCCN low-risk criteria (cT1c-T2a and biopsy Gleason grade group I and/or PSA < 10 ng/ml).
We excluded patients with number of biopsy cores < 10 or > 14, as well as cases with missing information (PSA, clinical T stage, biopsy Gleason grade group, number of positive prostate biopsy cores). This resulted in a cohort of 27,901 patients exposed to any treatment modalities: radical prostatectomy (RP), radiotherapy (RT) and non-local treatment (NLT). The main study group in whom upgrading and/or upstaging rates were examined consisted of 9355 RP patients. Here, we excluded those individuals with missing information on pathologic T stage and Gleason grade group (GGG). This resulted in 9126 assessable patients.

Statistical analyses
In the first part of the analyses, patient characteristics were compared across treatment modalities (RP, RT, NLT). In the second part, we reported rates of upgrading and/or upstaging at RP. Upgrading was defined as change from GGG1 at biopsy to GGG ≥ 2 at RP. A stricter definition of upgrading relied on change from GGG1 at biopsy to GGG ≥ 3 at RP. Upstaging to nonorgan-confined stage (NOC) was defined as presence of extracapsular extension of the tumor (ECE, pT3a) or seminal vesicle invasion (SVI, pT3b) or pT4/pN1 stage at RP. Additionally, different combinations of upgrading and/or upstaging were computed and survival outcomes for these combinations were recorded. Moreover, we stratified upgrading and/or upstaging rates according to PSA (≤ 4.5 vs 4.5-5.8 vs > 5.8 ng/ml), percentage of positive cores (≤ 17 vs 17-33 vs > 33%), number of positive cores (≤ 2 vs 3-4 vs ≥ 5), age (< 60 vs > 60 years ) and race/ethnicity (Caucasians vs African-Americans vs Hispanic vs Asian vs others). Last, but not least, multivariable logistic regression models predicting two definitions of combined upgrading and/or upstaging were fitted and the accuracy of the models was calculated with previous methodology 19,20 . All tests were two-sided with a level of significance set at p < 0.05 and R software environment for statistical computing and graphics (version 3.4.3) was used for all analyses.

Upgrading and upstaging rates in low-risk prostate cancer patients
Overall (  Fig. 1. Finally, detailed upgrading and/or upstaging rates according to patients characteristics are also tabulated (Supplementary Table 1).   In model 1, age, PSA, clinical T stage, percentage of positive cores, number of positive cores, but not race/ethnicity were independent predictors of pT2 with GGG2 ( Table 3). The accuracy of the model was 63.1% (95%CI 62.0-64.2%). In model 2, age, PSA, percentage of positive cores and number of positive cores, but not clinical T stage and race/ethnicity were independent predictors of NOC and/or GGG ≥ 3 ( Table 3). The accuracy of the model was 63.9% (95%CI 62.3-65.6%).

Survival differences in low-risk PCa patients according to pathologic stage and grade at RP
Overall, 12 deaths from PCa vs 139 deaths from other causes were recorded during 38,476 person-years of follow-up. After stratification according to different combinations of upgrading and/or upstaging definitions (NOC with GGG ≥ 3; NOC with GGG2; NOC with GGG1; pT2 with GGG ≥ 3; pT2 with GGG1-2), the following survival outcomes were recorded (Table 4).
In NOC with GGG ≥ 3 (n = 149), one death from PCa vs four deaths from other causes were recorded during 610 person-years of follow-up. This resulted in 1.6% persontime PCa-specific death rate per 1000 person-years. In NOC with GGG2 (n = 573), 1 death from PCa vs 11 deaths from other causes were recorded during 2404 person-years of follow-up. This resulted in 0.4% person-time PCa-specific death rate per 1000 person-years. In NOC with GGG1 (n = 233), one death from PCa vs four4 deaths from other causes were recorded during 1,018 person-years of followup. This resulted in 1.0% person-time PCa-specific death rate per 1000 person-years. In pT2 with GGG ≥ 3 (n = 397), one death from PCa vs three3 deaths from other causes were recorded during 1631 person-years of follow-up. This resulted in 0.6% person-time rates of PCa-specific deaths per 1000 person-years. Finally, in pT2 with GGG1-2 (n = 7774), 8 deaths from PCa vs 117 deaths from other causes were recorded during 32,814 person-years of follow-up. This resulted in 0.2% person-time PCa-specific death rate per 1000 person-years.

Discussion
We hypothesized that contemporary low-risk patients harbor lower rates of upgrading and/or upstaging than their historical counterparts due to important differences in diagnostics. We tested this hypothesis within a large contemporary cohort of low-risk PCa patients who underwent RP. Additionally we compared baseline characteristics of low-risk RP patients to their low-risk RT and NLT counterparts, to ensure absence of clinically meaningful differences.
First, regarding differences between low-risk RP vs RT vs NLT patients, RP patients harbored more aggressive PCa phenotype than their RT or NLT counterparts. Specifically, RP patients harbored a highest percentage of positive cores than RT and NLT patients (median, 25 vs 17 vs 14%, p < 0.001). Moreover, RP patients harbored a higher proportions of cT2a stage than NLT (7.1 vs 5.4%), but not RT (7.1 vs 7.1%) patients. In consequence, based on more unfavorable baseline PCa characteristics, it may be postulated that upgrading and/or upstaging rates recorded in RP patients, might be higher than those affecting RT or NLT patients. Unfortunately, this hypothesis cannot be formally tested since pathological staging only applied to RP patients and true PCa stage is unknown in those managed with RT and/or NLT.
Second, we identified low, albeit clinically meaningful rates of upgrading and/or upstaging. Upstaging from clinically favorable risk to pathologic NOC PCa represents the most important rate limiting factor, when conservative management strategy is implemented. Of 9126 low-risk PCa patients, 955 (10.5%) exhibited NOC stage. The largest subgroup consisted of pT3a (8.7%; 796/9126), . Additional 573 (6.3%; 573/9126) individuals will be misclassified with less far-reaching implications, based on presence of exclusive secondary Gleason pattern 4 (GGG2) and NOC. This small patient group (1.6% plus 6.3%) represents the fatal flaw of low-risk group assignment based on PSA, clinical T stage and biopsy alone. This rate may possibly be reduced significantly with the use of MRI 21,22 . Third, we tested whether upgrading and/or upstaging can be accurately predicted. Despite our best attempts and use of usual predictive modeling techniques the accuracy of predictions was low for both endpoints: 63.1% for NOC and/ or GGG ≥ 2 and 63.9% for NOC and/or GGG ≥ 3. In consequence, observed rates of upgrading and/or upstaging cannot be accurately predicted using the available risk factors, at least in the current study. It may be argued that introduction of MRI findings may significantly increase the predictive ability of such a model. To the best of our knowledge, two nomograms relied on MRI findings in addition to clinical variables to predict NOC and/or GGG ≥ 3 in European 19 and North American 25 cohorts of low and intermediate risk PCa patients. All two nomograms exhibited an added benefit, when MRI findings were added to clinical variables. However, these results need to be replicated in external validation samples. Taken together, since the ability of MRI to improve prediction of upgrading and/or upstaging is still under debate 26 , other clinical tools, such as PSMA PET/CT scanning 27 , biomarkers 28 (the prostate health index, urine PCA3, 4Kscore, TMPRSS2-ERG and ConfirmMDX) and IHC 29 and genetic risk score 30 may prove of value. However, large-scale contemporary testing of these tools have not been completed, especially in independent external validation samples that include sufficient numbers of low-risk patients 31 .
Fourth, we examined survival outcomes in the overall cohort of RP low-risk PCa patients with a median followup of 54 months. Expectedly, we recorded extremely low PCa mortality rates in these patients during 38,476 personyears of follow-up. Only 12 deaths were attributable to PCa. PCa-specific deaths ranged from 1 to 8, after stratification according to different combinations of upgrading and/or upstaging definitions. Based on person-time PCa-specific death rate per 1,000 person-years, only NOC with GGG ≥ 3 (1.6%) distinguished itself from values recorded for other combinations, including NOC with GGG2 (0.4%), NOC with GGG1 (1.0%), and pT2 with GGG ≥ 3 (0.6%). Nonetheless, all these combinations of upgrading and/or upstaging definitions invariably exhibited higher person-time PCa death rate for 1,000 person-years than pT2 GGG1-2 patients (0.2%) that represents the ideal candidate for AS. However, despite adequate median follow-up (54 month), the extremely low absolute number of PCa deaths undermined the ability to perform usual cancer-specific mortality rate comparisons. In consequence, consideration of person-years of follow-up and comparison of PCa person-time PCa death rates for 1,000 person-years is of essence. In agreement with our findings, Brooks et al. observed that adverse pathology (NOC and/or GGG ≥ 3) at the time of RP is highly associated with future development of distant metastasis and PCaspecific death among low and intermediate risk PCa patients from a single institutional prospectively maintained database (N = 428) with extensive follow-up (20 years) 32 .
The current study is not devoid limitations. The first and foremost limitation of the current study relates to the type of patients included in the analyses of upgrading and/or upstaging. Since biopsy required comparison to RP pathology, invariably such individuals represent RP patients. In consequence our findings are only directly applicable and mostly generalizable to similar patients. To address this limitation, we compared low-risk RP cohort to low-risk RT and NLT patients, with respect to age, race-ethnicity and clinical PCa characteristics. In those comparisons low-risk RP patients exhibited younger age and worse PCa characteristics. In consequence, RP low-risk PCa patients, at least in the current study, most likely reflect a less favorable phenotype of lowrisk PCa. Second, an additional important limitation is lack of MRI data, since MRI represents a standard of care in the diagnostic workup of not only newly diagnosed PCa patients, but also of individuals at risk of PCa 2,4 . Last but not least, SEER represents a retrospective data repository. Although, no prospective trial addressed the end points of the current study, several large institutional databases relied on prospective gathered data on biopsy and pathologic findings 33 . Such databases are undoubtedly better. However, they do not offer the same sample size of low-risk PCa patients as SEER.

Conclusions
In low-risk PCa patients, critical changes between tumor grade and stage at biopsy vs RP may be expected in few patients: NOC with GGG ≥ 3 in 1.6% and NOC with GGG2 in 6.3%. Others patients with upgrading and/or upstaging combinations will invariably harbor either pT2 or GGG1 that far less critically affect PCa prognosis.
Funding Open access funding provided by Università degli Studi di Roma La Sapienza within the CRUI-CARE Agreement. The research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Data availability statement All data generated for this analysis were from the SEER database (https:// seer. cancer. gov/ data/). The code for the analyses will be made available upon request.

Declarations
Conflict of interest The authors declare that there is no conflict of interests.
Ethics consent statement All analyses and their reporting followed the SEER reporting guidelines. Due to the anonymously coded design of the SEER database, study-specific institutional review board ethics approval was not required.

Patient consent statement, permission to reproduce material from other sources and clinical trial registration Not applicable.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.