Prostatic Ductal Adenocarcinoma Controlled for Tumor Grade, Stage, and Margin Status Does Not Independently Influence the Likelihood of Biochemical Recurrence in Localized Prostate Cancer After Radical Prostatectomy

After obtaining institutional review board approval, the first author reviewed 1631 consecutive RPs from January 2014 to May 2020 performed at the University of Miami, Miami, Florida. Most of the prostatectomies were performed by the 2 senior authors. Prostate glands were weighed without seminal vesicles, inked, fixed overnight in formalin, cut horizontally at 3-mm intervals, and entirely submitted for histologic examination.¹⁷  Prostate-specific antigen (PSA) density was calculated by dividing the preoperative serum PSA level by prostate weight without seminal vesicles.¹⁸  The base and apex of the prostate were shaved, serially sectioned, and submitted as perpendicular sections for margin assessment. Bilateral seminal vesicles were submitted for histologic examination. Each tumor nodule (TN) was mapped on histologic slides, graded individually, and staged.^16,18  Foci of mucinous prostatic adenocarcinoma and foamy gland prostatic adenocarcinoma were graded based on the underlying architecture. Prostatic ductal adenocarcinoma was determined by the presence of true fibrovascular cores lined by high-grade cells with elongated pseudostratified nuclei (Figure 1). The percentage of PDA in the TN was recorded. Prostatic ductal adenocarcinoma without necrosis was graded as Gleason pattern 4. Prostatic ductal adenocarcinoma with comedonecrosis was graded as Gleason pattern 5. The presence of cribriform morphology in invasive cancer, regardless of size and including glomeruloid structures, was recorded. Intraductal prostatic adenocarcinoma was not graded and not included in the TN Grade Group (GG)/Gleason score composition.^7,19,20  In cases where the distinction between invasive cribriform prostatic adenocarcinoma and intraductal prostatic adenocarcinoma could not be made based upon morphologic criteria alone and such distinction would not change the overall TN grade, no basal cell immunohistochemical analysis was performed, and the cribriform glands without comedonecrosis were considered Gleason pattern 4.²¹ 

Tumor nodules were considered spatially separate if they were 3 mm or more apart in the same plane of section or 4 mm or more apart on consecutive serial sections.¹⁸  To determine tumor volume in each nodule, slides were photocopied in a background of a 1-mm² grid and the number of square millimeters in each TN was manually counted. To convert square millimeters into cubic millimeters, the total number of square millimeters was multiplied by 3 (thickness of prostate tissue sections) and 1.12 (fixation shrinkage factor).²²  The carcinoma was considered as extending to the surgical resection margin when tumor was present at the inked cauterized margin.²³  We selected all cases with localized prostate cancer (ie, those lacking regional lymph node and distant metastasis) where the TN with the highest stage and grade and most adverse margin status contained PDA (ie, the cases in which all adverse RP findings were observed in the same dominant TN containing PDA). Discordant cases that had either PDA in secondary TNs or secondary TNs that demonstrated adverse pathologic RP findings (eg, positive surgical margin status) were excluded. We then selected for stage-, grade-, and margin status–matched control cases demonstrating classic PAA operated within a 1-year period from the reference surgical cases. Biochemical recurrence was defined as serum PSA 0.2 ng/mL or more on 2 separate readings. Patients who never achieved biochemical remission or those without follow-up were excluded from the study. Dates of BCR, death, or latest follow-up were used as observation end points.

Categorical data were compared by the Pearson χ² test. The Mann-Whitney U test and Welch 2-sample t test were used to assess differences in means of continuous variables. Biochemical recurrence–free survival was assessed by the Kaplan-Meier method and log-rank test. A multivariable Cox regression model was used to evaluate the effects of cancer grade, presence of PDA, its percentage, patient age, PSA density, stage, tumor volume, cribriform morphology, intraductal prostatic adenocarcinoma, and surgical margin status on the likelihood of BCR.

We reviewed a total of 1631 consecutive RPs performed at our institution between January 2014 and May 2020. Of these, we excluded 39 RPs that were performed following neoadjuvant hormonal and/or radiation therapy, 5 cases of vanishing prostatic carcinoma,²⁴  1 case of peripheral gland adenosis,²⁵  1 carcinoid tumor, and 1 small cell carcinoma. The remaining 1584 cases (which included 264 GG1 [16.7%], 620 GG2 [39.1%], 272 GG3 [17.2%], 50 GG4 [3.2%], and 378 GG5 [23.8%]) demonstrated treatment-naive prostate cancer from which PDA cases and corresponding controls were selected.

Prostatic ductal adenocarcinoma was present in the dominant TN, determined by the highest grade, stage, and margin status, in 147 cases, which included 26 GG2 (17.7%; Gleason score 3 + 4 = 7), 44 GG3 (29.9%; Gleason score 4 + 3 = 7), 6 GG4 (4.1%; Gleason score 4 + 4 = 8), and 71 GG5 (48.3%; 63 Gleason score 4 + 5 = 9 and 8 Gleason score 5 + 4 = 9) tumors. One TN was composed of pure PDA. In other cases, the proportion of PDA ranged from 5% to 90% (median, 20%). In 2 cases, PDA had comedonecrosis, and those areas were classified as Gleason pattern 5. Except for one case with pure PDA, there were no TNs in which the high-grade component was only PDA. In one case, the surgical margin was positive for carcinoma in the areas of EPE in both the dominant GG5 (Gleason score 4 + 5 = 9) TN featuring PDA and a secondary GG2 (Gleason score 3 + 4 = 7) TN without PDA. One case had PDA in 2 organ-confined TNs: a smaller GG3 (Gleason score 4 + 3 = 7) TN and a larger GG2 (Gleason score 3 + 4 = 7) TN. Seventy-six cases had EPE (pT3a), 33 cases had seminal vesicle invasion (pT3b), and 65 cases had a positive surgical margin, including 7 cases with a positive margin in the area of intraprostatic incision and no EPE elsewhere in the gland (pT2+). These 147 cases were matched with acinar adenocarcinoma cases based on the above-described criteria.

Twenty-four cases with PDA did not satisfy our inclusion criteria. One case had a smaller organ-confined GG4 (Gleason score 4 + 4 = 8) TN with PDA and a larger GG2 (Gleason score 3 + 4 = 7) TN without PDA that had a positive surgical margin in the area of EPE. One case had a larger organ-confined GG3 (Gleason score 4 + 3 = 7) TN with PDA and a smaller GG2 (Gleason score 3 + 4 = 7) TN without PDA that had a positive surgical margin in the area of EPE. Nine cases had PDA in the TN that was not the highest grade per case. Thirteen cases with PDA had metastasis in regional lymph nodes and could not be considered localized disease. Five of these were GG3 (Gleason score 4 + 3 = 7), including 3 with a minor higher-grade component (tertiary pattern 5), and 8 were GG5.

Follow-up was available for 126 PDAs and 118 classic PAAs. Thirteen men with PDA and 9 men with PAA never achieved a biochemical remission after RP, leaving 113 ductal and 109 acinar adenocarcinoma cases for final analysis (Table 1). When we compared these cohorts, men with PDA were on average 2 years older (P = .01), and their average tumor volume was roughly larger by a third than that of acinar cancers (P = .03). Other features did not differ significantly between the 2 matched cohorts. One man with PAA died of a cause not related to prostate cancer, and no prostate cancer–specific death was observed. The median time to BCR was 4 months for PDA and 6 months for PAA (P = .40). The men who did not experience BCR were followed for an average of 24 (PDA) and 28 (PAA) months (P = .50).

The results of multivariable Cox regression analysis are presented in Table 2. Prostate-specific antigen density (odds ratio, 3.7; P = .001), cancer grade (odds ratios, 3.2–4.3; P = .02), surgical margin status (odds ratio, 1.7; P = .02), and tumor volume (odds ratio, 1.3; P = .02) were significant independent variables associated with BCR on multivariable analysis. The presence of PDA, its proportion, cribriform Gleason pattern 4, and patient age were not independently associated with BCR (all P > .05). Factoring the effects of significant variables, the mere presence of PDA was not an independent predictor of BCR after RP (Figure 2).

Prostatic ductal adenocarcinoma is a less frequent variant of prostate cancer that has been historically considered a more aggressive subtype.^8–14  Although it was first described in 1967, grading of PDA was not officially addressed until the 2005 International Society of Urological Pathology consensus grading conference, where it was agreed that PDA without necrosis warranted a Gleason pattern 4 and that PDA with comedonecrosis should be assigned a Gleason pattern 5.⁶  However, in a recent study by Iakymenko et al,¹⁶  it was shown that once tumor grade and volume are accounted for, PDA does not have an independent influence on adverse pathologic findings at the time of RP. In the current study, we investigated the association of PDA with the likelihood of BCR after RP, controlling for other potential confounding factors that are known to be associated with worse clinical outcomes (ie, tumor grade, pathologic stage, and margin status).

Pure PDA is quite rare, with a reported incidence of up to 1.3% in some series.²⁶  In its more frequent presentation, admixed with classic PAA, the reported incidence is in the range of 2.6% to 13.4% of cases.^{5,12,13,15,27}  In our study, 171 cases had at least 1 TN with PDA, comprising 10.8% of 1584 consecutive RPs with treatment-naive prostate cancer. The frequency of PDA at RP has been increasing because of contemporary use of active surveillance for men with low-risk prostate cancer. Although our historical RP cohort included 16.7% GG1 and 39.1% GG2 cancers, only 16.6% of RP specimens with PDA were GG2. Thus, a change in prostate cancer management and increased use of surgery for high-risk disease where PDA is more likely to be present may account for its increased incidence observed in contemporary cohorts. In addition, cases with PDA demonstrated a higher average tumor volume, a significantly higher incidence of locally advanced disease, and positive surgical margins. Many of these factors have been proven by large-scale studies^28–33  to be independent significant variables associated with BCR and prostate cancer–specific mortality.

When PDA was matched with equally advanced classic PAA, PDA tended to have a lower PSA density despite having a higher average tumor volume. As shown by Morgan et al,¹¹  PDA can have a lower PSA production compared with PAA, which explains why the difference in tumor volume did not lead to a consequential difference in PSA density as its indirect measure in our cohort. We do not think that a lower PSA production by PDA interferes with our study design because only 1 man had pure PDA in the dominant TN. This patient is being followed by our team, and thus far he is clinically free of disease.

The presence of PDA had no independent effect on BCR. However, PSA density, tumor grade, tumor volume, and margin status were all independent predictive variables for BCR. The mere presence of PDA may not explain the increased likelihood of BCR. Instead, the presence of adverse pathologic findings typically found in cases with advanced high-grade locoregional disease (and in cases in which PDA is usually seen) may be more responsible for worse prognosis after surgery. A similar explanation can be offered for why the presence of intraductal carcinoma and cribriform Gleason pattern 4 did not independently influence the likelihood of BCR in our cohort. Such a change in the perception of the clinical significance of PDA variant morphology is not unique. For example, the presence of glandular or squamous differentiation in urothelial carcinoma was historically believed to be an adverse prognostic finding.³⁴  However, a study by Mitra et al³⁵  with a similar design to our current study demonstrated that, in a cohort of patients with divergent (ie, glandular or squamous) differentiation at cystectomy compared with a stage- and grade-matched cohort with urothelial carcinoma, the mere presence of divergent differentiation did not alter the postoperative course. Subsequent studies have arrived at similar conclusions that have been endorsed by the World Health Organization.³⁶ 

A population-based study by Meeks et al¹⁰  identified 693 men in the Surveillance, Epidemiology, and End Results database diagnosed with pure PDA. Although this study had a large number of cases, the specimens combined RPs and transurethral resections and included a cohort from a time period between 1970 and 2007 when both recognition and grading of PDA may have had low reproducibility, further complicated by inherent limitations of this database (ie, nonstandardized specimen submission and interpretation, lack of central pathology review).³⁷  Morgan et al¹¹  used the same database for a similar time period and arrived at the conclusion that PDA is more likely to present with advanced stage and has a higher likelihood of prostate cancer–specific mortality. This study highlights potential limitations of using a national database to draw conclusions based on pathologic findings, because there were cases of PDA recorded as Gleason score 6 or less.

There are few studies with contemporary pathology review that have investigated the impact of PDA on clinical outcomes following RP. Flood et al⁸  examined a cohort treated at Ottawa Hospital in which urologic pathologists rereviewed cases identified as GG5 prostate cancer. Although the significance of PDA was not the primary goal of the study, the authors concluded in univariable analysis that it was associated with BCR among GG5 cancers. The authors also observed that an increasing number of Gleason pattern 5 morphologies showed a significantly higher risk of BCR, but no detailed results of the independent effect of PDA on BCR were provided. At least 2 studies have investigated the association between percentage of PDA in RP specimens and clinical outcomes. Samaratunga et al¹²  concluded that the mere presence of PDA, rather than the proportion of this finding, was associated with an increased likelihood of pT3 disease. This conclusion is similar to our current and prior observations, where we recorded the presence of PDA in usually more locally advanced cancers.¹⁶  A study by Jang et al¹⁵  assessed the effect of the proportion of PDA at RP on the likelihood of BCR. The authors reported differences in BCR-free survival for patients with classic PAA, cases with less than 30% PDA, and cases with more than 30% PDA. Increasing tumor grade, stage, and incidence of positive surgical margins that paralleled an increasing percentage of PDA were associated with differences in outcomes and determined worse prognosis. In our cohort, the percentage of PDA did not independently influence the likelihood of BCR once we controlled for these other significant variables.

Ranasinghe et al³⁸  recently conducted a single-center retrospective review of 163 men with PDA and 155 men with high-risk classic PAA treated with RP. The authors reported no difference in the incidence of BCR between PDA and acinar cancers (77 of 163 versus 57 of 155; P = .069). But in multivariable analysis, they demonstrated that men with PDA had worse metastasis-free and overall survival after RP. Metastasis-free survival of pure PDA was described as worse compared with mixed acinar and ductal cancer, as well as pure acinar cancer. Ranasinghe et al³⁸  grouped pT3 and pT4 cases as well as GG4 and GG5 cancers for analysis, which may explain some of the differences observed between their study and our findings. Furthermore, data regarding specimen processing (ie, entire or partial gland submission for histologic examination) and grading (ie, global grading versus grading of individual TNs) were not provided for comparison and may also influence reported outcomes.

Our study has significant strengths, including a homogeneous surgical approach, submission of entire RP specimens for histologic examination, and central rereview of the entire RP cohort from which study and control cases were selected. Moreover, this study represents the largest centrally pathologically verified cohort of prostate cancer with ductal features. Patients who did not experience BCR had a median of ≥2 years of follow-up.

Men with PDA experience a significant incidence of BCR after RP. However, this observation appears to be more related to the presence of other adverse pathologic findings such as high tumor grade, advanced locoregional stage, and positive surgical margin status. Prostatic ductal adenocarcinoma did not demonstrate an independent association with BCR in localized prostate cancer when we controlled for these other independently significant variables. The presence of PDA preoperatively or in RPs without other adverse findings may not necessarily indicate a worse prognosis per se and should be considered in the context of other clinicopathologic factors for optimal disease management and patient counseling.

The authors thank Mr Anthonay De La Cruz and Ms Patricia Malinowski from the Department of Pathology and Laboratory Medicine for their outstanding administrative support for this study.