Old men with prostate cancer have higher risk of Gleason score upgrading and pathological upstaging after initial diagnosis: a systematic review and meta-analysis

To evaluate the predictive performance of age for the risk of Gleason score change and pathologic upstaging. Ovid MEDLINE, Ovid Embase, and the Cochrane Library were searched from inception until May 2020. Quality of included studies was appraised utilizing the Newcastle-Ottawa Quality Assessment Scale for case-control studies. The publication bias was evaluated by funnel plots and Egger’s tests. Our search yielded 27 studies with moderate-to-high quality including 84296 patients with mean age of 62.1 years. From biopsy to prostatectomy, upgrading and upstaging occurred in 32.3% and 9.8% of patients, respectively. Upgrading from diagnostic biopsy to confirmatory biopsy was found in 16.8%. Older age was associated with a significant increased risk of upgrading (OR 1.04, 95% CI 1.03–1.05), and similar direction of effect was found in studies focused on upgrading from diagnostic biopsy to confirmatory biopsy (OR 1.06, 95% CI 1.04–1.08). For pathologic upstaging within older men compared with younger, the pooled odds was 1.03 (95% CI 1.01–1.04). Thorough consideration of age in the context of effect sizes for other factors not only prompts more accurate risk stratification but also helps providers to select optimal therapies for patients with prostate cancer.


Background
Increasing prostate cancer (PCa) morbidity and mortality over past decades prompted the screening for general population and early detection for high-risk groups, as well as the improvement of manifold treatment modalities. PCa represents a spectrum of diseases that range from aggressive diseases that may require treatment to non-aggressive diseases that may do not. It is universally acknowledged that overdiagnosis and overtreatment could be averted by rational use of active surveillance (AS) and observation. Within younger informed men with long expectancy of life and favorable health status, patients diagnosed with low-to-intermediate-risk PCa may be considered for AS [1]. Compared to immediate curative treatment, AS delays the onset of potential side effects and is not associated with increased PCa mortality [2]. As for older patients with short expectancy of life and unfavorable health status, observation is recommended in order to maintain the quality of life and avoid the adverse events of unnecessary therapies [1]. Within this patient group, the survival rate of deferred treatment was also similar to that of immediate treatment [3]. Therefore, age is a pivotal factor for PCa in both risk appraisal and decision-making of treatment.
At initial diagnosis, clinical stage and Gleason score (GS) are both critical factors which provide urologists with the clinical profile of PCa to discriminate between indolent and aggressive disease. However, as diagnostic tools, both imaging tests and biopsy examinations inevitably have limitations in reflecting the tumor nature. Even if Gleason grading system has been modified over time, the accuracy of biopsy GS for predicting prostatectomy GS was reported to be barely satisfactory (53-74%) [4]. Pathological upstaging (hereafter referred to as upstaging) to more aggressive diseases was also reported to affect 7.2-17.2% of individuals [5,6]. It has been demonstrated in large-scale studies that GS upgrading (hereafter referred to as upgrading) and upstaging were significantly associated with biochemical recurrence, distant metastasis and death from PCa [7,8], while GS downgrading (hereafter referred to as downgrading) was a protective factor [9]. Various predictors such as prostate-specific antigen (PSA), prostate volume (PV), and number of biopsy cores were found to be associated with upgrading, downgrading or upstaging [10]. Age was also found to be predictive; however, value of odds ratio was slightly greater or less in some studies [11,12]. Furthermore, most studies were based on monocentric databases and limited population. To address these matters, we aimed to conduct a comprehensive review of literatures and investigate the correlation between age and the risk of GS change (upgrading or downgrading) and upstaging in this systematic review and meta-analysis.

Evidence acquisition
Our review was preformed according to the Preferred Reporting Items for Systematic Reviews and Metaanalysis (PRISMA) guideline [13]. Methods and criteria were specified in advance and documented in a protocol as a reference for our investigators.

Search strategy
A comprehensive literature search for eligible records was conducted via the following databases from inception until 5th May 2020: Ovid MEDLINE, Ovid Embase, and the Cochrane Library. The full PubMed search strategy was shown in Additional Figure 1. Search terms combined MeSH major topic with free-words. The major terms included 'Prostatic Neoplasms,' 'Multivariate Analysis,' 'Odds Ratio,' 'Disease Progression.' Free-words included 'upgrad*,' downgrad*,' 'upstag*,' etc. which were searched in all fields. The literature search was language-restricted to English and no article-type and publication-year restrictions were included in the search. Supplementary records were identified through reviewing references of relevant literatures.

Study criteria and selection
Studies were eligible if they met the following criteria: (1) original research with experimental designs; (2) studies in the research field of PCa; (3) studies focused on the association between age and GS change or upstaging; (4) statistical methods of multivariate analysis (MVA). Major exclusive criteria included (1) reviews, metaanalysis or comments; (2) studies with unavailable fulltext such as conference abstract; (3) studies of sample size fewer than 50 patients. See Fig. 1 for the PRISMA flow diagram detailing the criteria.

Study selection
Records were retrieved from electronic databases and citations identified from relevant literatures. Obtained records were deduplicated, and the remaining were screened via titles and abstracts for full-text review. If there existed studies reporting the overlapping results (from same database and same study period), we selected the one with largest sample size. The final included literatures were evaluated in both qualitative synthesis and quantitative synthesis (meta-analysis). This whole selection procedure was conducted by two investigators (XW, YZ) independently, and disagreements were resolved by consensus and approved by a third investigator (ZJ).

Data extraction
Data were independently extracted by two investigators (XW, YZ) from included literatures and any discrepancies were resolved by consensus and approved by a third investigator (ZJ). Procedures of extraction were performed using a standardized form including study characteristics (first author, year of publication, region and interval of study, study size, and definition of endpoints), patient information (age, PSA, PV, GS, clinical T-stage, and number of cores obtained) and statistical data (adjusted odds ratio (aOR) with respective 95% confidence interval (CI)) (shown in Tables 1 and 2).
In order to facilitate elucidation, studies were categorized into groups by definitions of study endpoints including (1) upgrading from biopsy to prostatectomy; (2) upgrading from diagnostic biopsy to confirmatory biopsy; (3) downgrading from biopsy to prostatectomy; (4) upstaging; and (5) upgrading and/or upstaging.

Risk of bias assessment
Two investigators (ZJ and QP) independently evaluated each included study utilizing the Newcastle-Ottawa Quality Assessment Scale (NOS) for case-control studies. Discrepancies in score assignment were later resolved by consensus.

Statistical analysis
The conversion to means of variables was roughly calculated using medians combined with interquartile range (IQR) according to Luo's methods [35]. Random effects models were used for all meta-analysis by computing log-transformed aOR (logaOR) with respective standard error (SE) to appraise the predictive performance of age on GS change and upstaging. Forest plots were depicted to provide the pooled results (i.e., the overall effect measure, Z statistic) and heterogeneity among studies. Heterogeneity was evaluated using I 2 statistic, which represented whether the variation was attributed to heterogeneity or chance. Further subgroup analysis and sensitivity analysis were carried out to find the possible sources of heterogeneity. The publication bias was evaluated by visually inspecting the asymmetry of funnel plots and subsequently quantified by Egger's tests. Tests of publication bias were only used when there were at least 10 studies included in each group otherwise power of tests were too low to distinguish chancer from real asymmetry. Tests were 2 sided and P = 0.05 was the threshold for statistical significance. Meta-analysis and statistical tests were performed using computer software of RevMan version 5.3 and Stata version 12.0.

Evidence synthesis
There were 2296 eligible records retrieved from electronic databases and additional 40 identified from pertinent references. Of retrieved records, 62 were deduplicated and the remaining 2274 were screened for fulltext review via titles and abstracts. Two hundred ninety articles were selected after screening, and 27 literatures were finally included according to exclusive criteria. The reasons for exclusion were provided in PRISMA flow diagram (shown in Fig. 1).

Study description
Included 27 literatures were all retrospective studies of case-control designs which were published between 2010 and 2020. Eighteen literatures were monocentric studies, 4 were multicentric, and 5 were from administrative databases. Total sample size of this review was 84,296 patients in a time span of 28 years (1990-2018). Irrespective of literatures based on administrative databases, sample size was 23,211 patients. There were 18 studies from North America, 7 from Europe, 4 from Asia, and 1 from Australia. Seventeen literatures were enrolled into group of upgrading from biopsy to prostatectomy with total 48,590 patients [5, 6, 10-12, 14-19, 21-25, 27], while 3 were enrolled into group of upgrading from diagnostic biopsy to confirmatory biopsy (n = 1778) [32][33][34]. Three literatures were eligible for group of downgrading from biopsy to prostatectomy (n = 13532) [5,9,29], 7 were eligible for group of upstaging (n = 22790) [5,6,10,11,16,19,28] and 5 were eligible for group of upgrading and/or upstaging (n = 20444) [5,20,26,30,31] (shown in Tables 1 and 2). From biopsy to prostatectomy, upgrading was found in 32.3% of patients with biopsy specimens, while downgrading was found in 19.0% of patients. Upgrading from diagnostic biopsy to confirmatory biopsy was found in 16.8%. Of 22,790 patients, the upstaging was found in 9.8% of organ confined diseases. Calculated with extractable data, the pooled mean age, PSA level, and PV were 62.1 years, 3.97 ng/mL, and 42.1 mL, respectively. Most patients had low-risk organ-confined PCa at diagnosis and underwent biopsy with extended schemes (shown in Tables 1 and 2)

Risk of bias assessment
All included literatures had low-to-moderate risk bias according to NOS scale. Three articles were rated as a total score of 5, while the others were rated as 6 or  PSA prostate-specific antigen, PV prostate volume, bGS biopsy Gleason score, pGS prostatectomy Gleason score, NA not available, IQR interquartile range a All studies were case-control designs more. Studies with data source of administrative databases instead of original records resulted in bias of selection and inaccuracy in ascertainment of exposure. Studies not adjusting for potential confounders in MVA resulted in poor comparability, especially the excluded confounders very likely influencing discordance of GS and disease stage between initial diagnosis and final outcomes (shown in Table 3)

Gleason score change
As shown in the forest plots (shown in Fig. 2), age was found to be an independent predictor of upgrading.

Heterogeneity analysis
In subgroup analysis and sensitivity analysis of upgrading (shown in Table 4), we managed to seek the source of heterogeneity. We found that I 2 statistic decreased to 67% when studies including patients of higher than cT2 were removed. Decreased heterogeneity (62%) was also found when studies not adjusted for PV were removed. Considering small size effect, I 2 statistic decreased to 44% and 12%, respectively in subgroups above after removed literatures of study size smaller than 300. After removing studies not adjusted for number of cores obtained, I 2 statistic decreased to 50%. Prostate size [10,17,18,21,22], number of cores [36], and disease stage [11,12,15] were validated to be independent predictors of upgrading. Existence of confounders was thought to be the major reason for heterogeneity of our synthesis. As for upstaging, after removed one literature [28] of small size, I 2 statistic decreased to 0%. Small size study made the pooled analysis unstable and less robust.

Discussion
In this systematic review and meta-analysis of 27 studies with moderate-to-high ranking of quality, we identified that age was an independent predictor for upgrading from biopsy to prostatectomy (aOR 1.04, 95% CI 1.03-1.05) and upstaging (aOR 1.03, 95% CI 1.01-1.04) in the patient groups who were eligible for curative treatment. For every increased year of age, there was a 4% increased risk of upgrading and 3% of upstaging. From diagnostic biopsy to confirmatory biopsy, there was a 6% increased risk of upgrading (aOR 1.06, 95% CI 1.04-1.08). Based on large-scale population, this review also demonstrated that upgrading, downgrading and upstaging occurred respectively in 32.3%, 19.0% and 9.8% of patients who underwent prostatectomy. Upgrading at confirmatory biopsy was found in 16.8% of initial biopsy. PCa is a spectrum of heterogeneous diseases, some of which are so indolent to influence overall survival especially in older patients. Therefore, precise assessment of risk stratification seems to be pivotal in disease management. It has been demonstrated that upgrading and upstaging were significantly associated with biochemical recurrence, unfavorable surgical outcomes, and cancerspecific survival [7,37,38]. Even if the proportion of both upgrading and upstaging is small, precise risk stratification and prediction of outcomes are requisite for individualized management.
For younger individuals especially with good health status, those with low or favorable intermediate risk PCa are recommended to undergo AS rather than immediate curative therapies [1]. Even if younger patients under AS had no significant difference in treatment-free survival compared with older counterparts [39], delayed treatment can avoid side effects such as erectile and voiding dysfunction in the young. According to the findings of this review, young patients had lower risk of upgrading and upstaging. Accurate assessment of PCa may boost confidence of urologists and patients in initiation and continuation of AS protocol. For patients with low risk of upgrading and upstaging who intend to undergo definitive treatment, salvage therapies might be unnecessary to improve prognosis. Longer waits between postoperative follow-ups might be recommended. Even if the risk is relatively low, confirmatory biopsy or rebiopsy is an option before developing and implementing PCa management programs.
For older asymptomatic individuals with other morbidities or short life expectancy, no further therapies are indicated until symptoms occurs [1]. Provided that older patients might have higher rate of upgrading and upstaging, monitoring the course of disease need to be performed in a relatively short space of time and the delivery of palliative therapies for the development of symptoms might be earlier. Older men had a significantly higher clinical stage and biopsy Gleason grade at diagnosis which was associated with poorer prognosis compared with younger men [40]. However, fewer older individuals with good quality of life might benefit from curative treatment. Thus, urologists should be reminded of high-grade PCa harbored in older patients and reconsider the treatment modalities before developing a treatment program. Included literatures all have different research purposes that focused on different variables which might impact on upgrading or upstaging. Laboratory findings [18,23] (such as neutrophil to lymphocyte ratios), imaging parameters [20] (such as PI-RADS score), pathological features [28] (such as number or proportion of positive biopsy cores) and other clinical factors [12] (such as years of research) were all investigated to have potential predictive performances. This might contribute to the heterogeneity of our review. Age is a basic parameter as well as a confounder which is usually adjusted for in MVA. And as such, age was validated to be an independent predictor under different circumstances.
In 2005 and 2014, amendments to the Gleason grading system by the International Society of Urological Pathology (ISUP) were recommended [41]. This has led to a closer correlation between the grade of a series of thin core biopsies and the subsequent radical prostatectomy specimen. It has been reported that Gleason score agreement improved and the proportion of biopsy cores upgraded decreased from before 2005 to after 2005 [42]. However, many studies have demonstrated that age was independent predictor even if it was adjusted for year [11,15,19,20,30]. Even so, further long-term studies should also consider the version of grading system as a confounder.
It is highly debatable whether PCa progresses during long interval from biopsy to prostatectomy [43,44]. It has been reported that actuarial risk of grade progression during 1-year follow-up was 3% among patients with Gleason 6 or less organ-confined tumors (11% for 2 years) [45]. Even if the majority of studies did not report the intervals, we assumed that variability of time did not influence the comparison among studies. However, it is worth noticing that different intervals could be the potential source of heterogeneity.
Except the heterogeneity discussed above, our review has several limitations. First of all, there was variability in study designs among our included studies in which only a few focused on the predictive performance of age. Secondly, inevitable reasons such as unavailable full-text and non-extractable data resulted in incomplete retrieval  of research. Moreover, most studies only consisted of population with low-risk PCa and focused mainly on disease reclassification from GS ≤ 6 to higher, which might not represent the reality. Last but not least, different biopsy protocols influence detection rates of low-grade or high-grade diseases [36,46]. Different schemes of biopsy and different imaging guidance which were not reported in some studies also contribute to significant variation in outcomes.

Conclusions
Age is an independent predictor for both Gleason score upgrading and pathological upstaging. Combined with other prognostic factors, thorough consideration of age not only prompts more accurate risk stratification but also helps providers to select optimal therapies for patients with prostate cancer. Nevertheless, further robust studies are necessitated to confirm these results in the context of effect sizes for other factors such as PSA, number of positive cores, and race.