Large cribriform growth pattern identifies ISUP grade 2 prostate cancer at high risk for recurrence and metastasis

Invasive cribriform and intraductal carcinoma are associated with adverse clinical outcome in patients with Gleason score 7 prostate cancer. It is yet unclear whether invasive cribriform and intraductal carcinoma of the prostate both have independent prognostic value, or whether field size of invasive cribriform carcinoma has impact on disease outcome. Our objective was to determine the prognostic impact of intraductal and invasive cribriform prostate cancer histological subtypes in radical prostatectomies. We reviewed 420 prostatectomy specimens with ISUP grade 2 prostate cancer, assessed the percentages of Gleason grade 4 and tertiary 5, and performed immunohistochemistry for basal cells to discriminate intraductal from invasive cribriform growth. Small and large invasive cribriform fields were distinguished based on a diameter of at least twice the size of adjacent pre-existent normal glands. Clinicopathological parameters and biochemical recurrence-free survival were used as endpoints. Cribriform architecture was observed in 228 (54.3%) men, 103 (24.5%) of whom had intraductal, 194 (46.2%) small invasive, and 34 (8.1%) large invasive cribriform growth. Large invasive cribriform architecture was associated with older age (P < 0.001), higher percentage Gleason grade 4 (P = 0.001), extraprostatic expansion (P < 0.001), and more frequent lymph node metastases (P = 0.002), when compared with small invasive cribriform and/or intraductal carcinoma. Univariate analysis identified PSA, pT-stage, surgical margin status, and intraductal and invasive cribriform growth as significant predictors for biochemical recurrence-free survival. In multivariable Cox regression analysis, pT-stage (hazard ratio = 1.64, 95% CI: 1.02–2.63, P = 0.04), positive surgical margins (hazard ratio = 3.28, 95% CI: 2.06–5.23, P < 0.001), and large cribriform growth (hazard ratio = 4.36, 95% CI: 2.08–9.17, P < 0.001) were independent predictors for biochemical recurrence-free survival, while intraductal carcinoma, small cribriform growth, and percentage of Gleason grade 4 were not. In conclusion, large cribriform fields represent an aggressive subpattern of invasive cribriform prostate cancer and are an independent predictive factor for biochemical recurrence-free survival in ISUP grade 2 prostate cancer patients.


Introduction
The Gleason score is one of the most important parameters for clinical decision-making in men with prostate cancer. The Gleason grading system is entirely based on tumor architectural growth patterns which are classified into five different grades. While men with biopsy Gleason score 6 are frequently eligible for active surveillance, treatment is warranted in patients with Gleason score 8- 10. The optimal therapeutic strategy for individual patients with Gleason score 7 is not yet clear. While most patients with Gleason score 7 undergo radical prostatectomy or radiation therapy, active surveillance is increasingly being considered in this large group of men. Therefore, there is an urgent need for additional parameters to aid therapeutic decision-making in men with Gleason score 7 prostate cancer.
Gleason score 7 prostate cancer is composed of welldelineated Gleason grade 3 glands along with Gleason grade 4 structures. Gleason grade 4 prostate cancer is heterogeneous, comprising a range of growth patterns, categorized as poorly formed, fused, glomeruloid, and cribriform [1,2]. These individual growth patterns are generally not specified in pathology reports, however, several studies have found that patients with invasive cribriform growth have a worse outcome than men without this pattern [3][4][5][6][7][8][9]. Among cribriform prostate cancers heterogeneity of architectural pattern is still present, with some areas being round and small, while others are large and confluent vastly exceeding pre-existent gland diameter [10,11].
Intraductal carcinoma of the prostate is characterized by either cribriform or solid malignant epithelial proliferation, or loose cribriform and micropapillary formations of severely atypical cells, in pre-existent large acini and prostatic ducts, with preservation of basal cells [2]. Although intraductal carcinoma is formally not included in the Gleason score, numerous studies have linked intraductal carcinoma to more aggressive disease [11][12][13][14][15][16]. The presence of intraductal carcinoma ought thus to be routinely noted in pathology reports [2,17].
Invasive cribriform Gleason grade 4 prostate cancer and intraductal carcinoma often coexist, and can be difficult to distinguish without the use of immunohistochemical staining of basal cells. At present, it is not clear whether invasive cribriform carcinoma and intraductal carcinoma both have independent prognostic value for prostate cancer, or whether invasive cribriform subpatterns have additional prognostic value [5,18]. The objective of this study is to determine the outcome of invasive cribriform subpatterns and intraductal carcinoma in patients with ISUP grade 2 after radical prostatectomy.

Patient selection
In total 854 patients were identified who had undergone radical prostatectomy for prostate adenocarcinoma at the Erasmus MC, Rotterdam, The Netherlands between 2000 and 2017. Men who had undergone hormonal, radiation or viral therapy (n = 19) prior to operation were excluded from this study [19]. After fixation in neutral-buffered formalin, the radical prostatectomy specimens were sectioned transversely and entirely embedded for diagnostic purposes. All slides and blocks were available for pathology review.

Pathologic evaluation
Two investigators blinded to clinical outcome (E.H. and Gv.L.) reviewed all radical prostatectomy specimens (n = 854). The following features were recorded: Gleason score according to the WHO/ISUP 2014 guidelines, pT-stage according to the AJCC TNM eighth edition, surgical margin status, presence of intraductal carcinoma, percentage Gleason grade 4, including specific growth patterns, and presence of tertiary Gleason grade 5 [2,20]. Based on this revision, ISUP grade 2 specimen were identified. The following Gleason grade 4 growth patterns were recognized: poorly formed, fused, glomeruloid, and cribriform glands [2,17]. In addition, we distinguished small and large cribriform growth patterns. Small cribriform structures had a diameter less than twice the size of adjacent benign glands. Large cribriform pattern was defined as having a diameter of at least twice the size of adjacent pre-existent normal glands, and could either represent one large well defined cribriform field or a confluent cribriform area ( Fig. 1). Invasive cribriform Gleason grade 4 was morphologically distinguished from intraductal carcinoma based on the following features: invasive cribriform prostate cancer had irregular outline, showed anastomosing fields beyond pre-existent gland architecture or extension into periprostatic fat tissue, ejaculatory ducts, or seminal vesicles. Intraductal carcinoma was morphologically identified if cribriform structures were clearly continuous with preexistent glands lined by normal basal epithelium, or containing corpora amylacea. Where invasive cribriform carcinoma and intraductal carcinoma could not be differentiated by morphological criteria alone, additional immunohistochemical staining for the presence of basal cells was performed; in total 261 slides from 156 ISUP grade 2 patients were stained. Gleason grade 5 was considered as a tertiary pattern if it occupied less than 5% of the total tumor area [1,2,17].

Immunohistochemistry
Four micrometer thick tissue sections were cut from selected paraffin-embedded blocks and mounted on slides (Superfrost Microscopic Slides, ThermoFisher Scientific, Bleiswijk). Slides were deparaffinized and rehydrated with xylene and ethanol. Endogenous peroxidase was blocked using 0.3% H 2 O 2 in phosphate-buffered saline. Heatinduced antigen retrieval was accomplished by 15 min in Tris-ethylenediaminetetraacetic acid buffer (pH 9; Klinipath, Duiven, The Netherlands). Mouse monoclonal highmolecular weight cytokeratin (clone 34BE12; 1:200; DAKO; Heverlee, Belgium) diluted in normal antibody diluent (APG-500; ScyTek Laboratories, West Logan, USA) was incubated for 2 h at room temperature. Antibody visualization was performed using the Envision kit (DAKO) and slide counterstaining with hematoxylin. When basal cell staining was completely absent around a cribriform gland, it was categorized as invasive cribriform carcinoma; if sporadic, scattered, or continuous basal cells were identified the structure was classified as intraductal carcinoma.

Clinical follow up
Clinical follow-up after radical prostatectomy consisted of six-monthly, and later annual monitoring of serum prostatespecific antigen (PSA) levels. Biochemical recurrence was defined as PSA level ≥0.2 ng/ml measured at two separate time points at least three months apart when PSA had been undetectable after operation, or as PSA increase of >2.0 ng/ ml if serum PSA had not declined to zero after operation. Postoperative lymph node and distant metastases were confirmed by biopsy or multidisciplinary consensus. Biochemical recurrence-free survival was defined as time in months from radical prostatectomy to biochemical recurrence.

Statistical analysis
Normally distributed, continuous variables were analyzed using the independent sample Student's t test, whereas variables without normal distribution were analyzed using the Mann-Whitney U test. Pearson's chi squared (χ 2 ) test was used for categorical parameters. Percentage Gleason grade 4 was analyzed both as continuous and dichotomous parameter (≥5% and <25% versus ≥25% and <50%).
Missing PSA values (n = 27) were imputated using the median PSA value. Biochemical recurrence-free survival was analyzed using Cox proportional hazards regression and visualized by Kaplan-Meier curves, excluding patients with lymph node metastases at time of operation. Statistics were performed using SPSS version 24 (IBM, Chicago, IL, USA). Results were considered significant when the twosided P value was <0.05.
In total, 228 (54.3%) patients showed either invasive or intraductal cribriform carcinoma. These patients had higher PSA levels (mean 12.2 ng/ml versus 9.4 ng/ml; P = 0.006) than those without cribriform architecture. They also more frequently had extraprostatic extension (51.8% versus 35.4%; P < 0.0001) and positive surgical margins (39.5% versus 27.1%; P = 0.007). One hundred and fifty (65.8%) patients with cribriform architecture and 91 (47.4%) patients without cribriform architecture had undergone pelvic lymph node dissection at the time of radical prostatectomy. Twelve of the (8.0%) patients with cribriform architecture were found to have lymph node metastasis at time of radical prostatectomy, compared to none in the group without cribriform architecture (P = 0.006).

Comparison of invasive cribriform and intraductal carcinoma
Detailed histopathological and immunohistochemical analysis revealed that invasive cribriform carcinoma was present in 204 (48.6%), and intraductal carcinoma in 103 (24.5%) cases. Solid and loose papillary morphological variants of intraductal carcinoma were rarely observed, and co-existed with cribriform intraductal carcinoma in each case. Seventy-nine (18.8%) men had both intraductal PSA levels (P = 0.06), pT stage (P = 0.32), surgical margin status (P = 0.36), and occurrence of lymph node metastasis (P = 0.39) were not statistically significant different between patients with invasive cribriform carcinoma without intraductal carcinoma (n = 125) and men with intraductal carcinoma only (n = 24). Patients with both invasive cribriform and intraductal carcinoma (n = 79) more frequently had extraprostatic extension (60.8% versus 46.4%; P = 0.02) and lymph node metastasis (11.4% versus 1.6%; P = 0.003) than those with invasive cribriform growth without intraductal carcinoma; there was no statistically significant difference in PSA level (P = 0.07), pT stage (P = 0.64), surgical margin status (P = 0.20), and lymph node metastasis (P = 0.32) between men with combined invasive cribriform and intraductal carcinoma and intraductal carcinoma only.

Large invasive cribriform carcinoma
Large invasive cribriform growth was observed in 34 (8.1%) patients. All of these men (100%) had concomitant small invasive cribriform growth and 24 (70.6%) had intraductal carcinoma (Table 1). We compared patients with invasive large cribriform growth with men who had either small invasive cribriform growth and/or intraductal carcinoma (n = 194

Clinical outcome of invasive and intraductal carcinoma
The median follow-up of ISUP grade 2 patients without positive lymph node dissection at time of radical prostatectomy (n = 408) was 53 months (interquartile range: 12.7-99.1 5.23, P < 0.0001), and invasive large cribriform architecture (hazard ratio = 4.36, 95% CI: 2.08-9.17, P = 0.0001) were independent predictors for biochemical recurrence-free survival, while small invasive cribriform growth pattern and intraductal carcinoma were not. To determine whether the difference in prognostic value between invasive small and large cribriform growth could be explained by an overall higher percentage of cribriform growth, we compared the outcome of patients with ≥5% invasive cribriform growth and those with <5%. When invasive cribriform growth was present, no statistically difference existed between low-and high-cribriform percentage (log rank; P = 0.087).
During follow-up 13 patients developed bone metastasis. Nine of these patients had small invasive cribriform or intraductal carcinoma (4.6%) and four had invasive large cribriform carcinoma (11.8%) at radical prostatectomy. The median time to bone metastasis was 138 months (interquartile range: 109.4-172.6) for small invasive and intraductal cribriform carcinoma and 59 months (interquartile range: 17.9-114.8) for invasive large cribriform carcinoma.
Due to the low number of events we were not able to perform further statistical analysis.

Discussion
While most patients with ISUP grade 2 prostate cancer are treated with radiotherapy and/or surgery, active surveillance is increasingly being considered as alternative strategy for these men [21][22][23][24][25]. Further risk stratification in this large group of patients is necessary to support therapeutic decision-making. Recently, invasive cribriform carcinoma and intraductal carcinoma have been recognized as promising additional predictive parameters for men with ISUP grade 2 prostate cancer [3-5, 11, 26]. In the current study, invasive and/or intraductal cribriform carcinoma was present in 54.3% of radical prostatectomies with ISUP grade 2 prostate cancer. While the clinicopathological features of men with invasive cribriform carcinoma without cribriform intraductal carcinoma were not statistically significant from those with cribriform intraductal carcinoma only, patients with both invasive and intraductal cribriform carcinoma more often had extraprostatic extension and lymph node metastasis than those with invasive cribriform carcinoma only. Furthermore, we found that patients with large invasive cribriform growth had higher pT-stage and more frequent positive lymph nodes than those with small invasive and/or intraductal cribriform carcinoma. In multivariable analysis, large invasive cribriform carcinoma was an independent predictor for biochemical recurrence-free survival, while small invasive carcinoma and intraductal cribriform carcinoma were not.
Various studies have addressed the association of either invasive cribriform carcinoma or intraductal carcinoma with adverse features at prostatectomy and with clinical outcome [3,5,8,11,27]. We observed that invasive and intraductal cribriform carcinoma were present in 48.6% and 24.5%, respectively of prostatectomy specimens with ISUP grade 2 prostate cancer. These rates are comparable to those found by others. Trudel et al. for instance found intraductal carcinoma in 17.5%, invasive cribriform carcinoma in 45.6%, and both invasive and intraductal cribriform carcinoma in 36.8% of 57 prostate specimen [11]. In a cohort of 286 ISUP grade 2 prostate cancer patients, Choy et al. demonstrated intraductal carcinoma in 26.5% and invasive cribriform growth in 38.7% [27]. Two studies took into account large cribriform architecture, however, these used different thresholds [6,11]. Iczkowski et al. defined large cribriform pattern as having more than 12 luminal spaces, while area size exceeding the size of an average benign gland was used by Trudel et al. Our threshold of large cribriform fields as at least twice the size of normal adjacent glands, exceeds that of previous studies. For instance, in the study of Iczkowski In a previous case-control study of 161 men with Gleason score 7 at radical prostatectomy, we found invasive cribriform but not intraductal carcinoma to be a significant predictive marker for metastasis-and disease specific-free survival in multivariate analysis [5]. In a subsequent analysis of prostate biopsies with long-term follow-up, both invasive and intraductal carcinoma had predictive value for disease-specific death, and combining both lesions had the strongest prognostic value [4]. The prognostic value of invasive and intraductal carcinomas at biopsies does not always correspond with the prognostic value at radical prostatectomies. Sampling artifacts inherently associated with diagnostic biopsies are likely the cause of discrepancies between biopsies and radical prostatectomy specimens. This is for instance reflected by the frequency of cribriform growth in biopsies and resection specimens; while invasive and/or intraductal cribriform architecture was found in 17% of sextant biopsies with ISUP grade 2, it was present in 54.3% of radical prostatectomy specimens in the current study [4]. Since most biopsy schedules currently include between 8 and 16 biopsies, and biopsies are increasingly being targeted by magnetic resonance imaging, the frequency of invasive cribriform, and/or intraductal carcinoma is higher with fewer sampling artifacts [28,29]. Since both small cribriform growth and intraductal carcinoma are often associated with large cribriform growth, these patterns should still be reported.
The outcome of this study may have important implications. First, we propose the inclusion of the presence of large invasive cribriform in pathology reports. Of 26 men with large cribriform architecture who had undergone pelvic lymph node dissection at time of radical prostatectomy, 23% (n = 6) had lymph node metastasis. Men with large cribriform architecture should therefore not be considered for surveillance but instead be offered active treatment with lymph node dissection. On the other hand, the absence of metastasis and low risk of biochemical recurrence in Gleason score 3 + 4 = 7 patients with no cribriform architecture might indicate that active surveillance can be considered in these men, and that pelvic lymph node dissection might be omitted when treatment is offered. However, it is important to note that the current results were obtained after studying radical prostatectomy specimens, while treatment decisions are made based on diagnostic biopsies. An urgent need exists to incorporate pathological features such as small and large invasive cribriform growth, as well as intraductal carcinoma, into clinical nomograms and prediction tools.
Strong points of this study are the detailed histological review and the extensive immunohistochemical staining for classification of cribriform architecture. Although large cribriform growth is an adverse predictive parameter for ISUP grade 2 prostate cancer patients, the stringent cut-off used in this study resulted in the inclusion of a relatively small number of cases and must be validated. Finally, the retrospective study design and relatively short median follow-up of 53 months possibly gave rise to a selection bias.
In conclusion, we demonstrate that patients with large invasive cribriform growth represent a more aggressive subgroup of cribriform ISUP grade 2 prostate cancer. Men with large invasive cribriform carcinoma should be actively treated since they are at increased risk for biochemical recurrence and metastasis.