High-Grade Desmoplastic Foamy Gland Prostatic Adenocarcinoma: A Study of 24 Cases

Twenty-four cases of high-grade desmoplastic foamy gland prostatic adenocarcinoma, sent in consultation to the senior author from 2010 to 2021, were analyzed. In all cases, these represented the patients’ initial biopsy. No patient received prior therapy. The vast majority were needle biopsies (n = 21), with 2 transurethral resection specimens, and a single cystoprostatectomy specimen obtained for urothelial carcinoma. Patient and clinicopathologic data were obtained from referring physicians.

In all cases, slides were submitted with basal cell markers (p63, high-molecular-weight cytokeratin [HMWCK]); if not done at the outside institution, we subsequently performed staining on either submitted or requested paraffin blocks/unstained slides. Immunohistochemical labeling was performed as previously described. Briefly, immunohistochemistry was done on the Benchmark XT autostainer (Ventana Medical Systems Inc, Tucson, Arizona) using I-View detection kit. The standard antibodies used, dilutions, pretreatments, and vendors were as follows: PTEN (1:100; No. 9188L, Cell Signaling), TP53 (predilute; No. 760-2542, Roche), and PIN4 (cocktail of 3 antibodies: CK903, predilute, 334M-88, Cell Marque; P63, predilute, PM163AA, Biocare Medical; Racemase, 1:200, Z2001RS, Zeta).

Clinical data are summarized Table 1. The 24 patients ranged in age from 40 to 89 years (mean, 67 years; median, 66 years). In 20 cases with available serum prostate-specific antigen (PSA) levels at the time of biopsy or resection, PSA levels ranged from 0.3 to 319.5 ng/mL (mean, 25.6 ng/mL; median, 7.1 ng/mL). Four patients had very low serum PSA levels (<1.0 ng/mL).

Of the 24 patients, 4 were lost to follow-up and 2 were diagnosed recently with only 1 to 2 months follow-up. Two patients had died at 1 and 6 years with no further information. Four patients had no evidence of disease with follow-ups of 7, 11, 12, and 15 months. The remaining 12 patients either developed metastases or had died of disease. Six men were alive with metastases. Two had bone metastases (alive at 50 months, 54 months). Two had lymph node metastases at the time of diagnosis (alive at 1 month, 2 months). One each had metastatic disease to the lung (alive at 13 months) and liver (alive at 6 months). Five men died of cancer with metastases to bone (dead at 12, 42, 45 months) or lung (23, 25 months). An additional patient had cancer that spread to the bone with no further information on whether he was dead or alive.

Overall, of the 16 patients with meaningful follow-up, 12 (75%) either had metastases or died of prostate cancer.

The defining feature of these tumors was a prominent desmoplastic reaction, observed in all cases. The dense desmoplastic stroma often obscured poorly formed glands and single cells (Figure 1, A through D). Another feature was that large glands with infolding often were disrupted in the stroma, resulting in difficulty in diagnosing malignancy and assigning a grade (Figure 1, E and F). Owing to the extensive desmoplastic stromal response, there was typically a pattern consisting of low tumor concentration with sparse solid sheets, with single or cords of tumor cells with large areas of intervening dense stroma, which caused further challenges in grading.

All cases had some morphologic feature of foamy gland carcinoma, with 7 cases (29%) showing prominent foamy gland morphology (Table 2). In 21 of 24 cases (88%), the predominant architectural pattern was cribriform carcinoma with and without necrosis (Figure 2, A and B). In 2 cases (8%) solid nests with and without necrosis predominated, and in 1 case (4%) the most common pattern was poorly formed glands (Figure 2, C and D). Other architectural patterns included large glands with infolding, single glands, cords of cells, and individual cells (Figure 2, E and F, and Figure 3, A through F). The radical cystoprostatectomy specimen had a sarcomatoid component with osteogenic sarcomatous differentiation and focal squamous cell carcinoma differentiation associated with the desmoplastic foamy gland carcinoma (Figure 4, A through D). In contrast to the bland cytology seen in usual foamy gland carcinoma, at least moderate nuclear atypia was observed in all the cases. There was greater cytologic atypia seen than in usual prostate adenocarcinoma in 11 of 24 cases (46%), yet not to the degree seen in pleomorphic giant cell adenocarcinoma of the prostate (Figure 5, A and B).

Associated intraductal adenocarcinoma was another typical feature of these tumors, observed in 22 of 24 cases (92%) with 4 of 24 cases (17%) demonstrating extensive intraductal adenocarcinoma (Figure 5, C and D). In 3 of 21 needle biopsy cases, there was usual limited Gleason score 3 + 3 = 6 (Grade Group 1) adenocarcinoma: 2 cases in cores separate from, and in 1 case in the same core as, the desmoplastic foamy gland carcinoma. In the cystoprostatectomy specimen there was usual Gleason score 3 + 4 = 7 (Grade Group 2) adenocarcinoma next to the desmoplastic foamy gland carcinoma.

Twenty of the 24 cases (83%) were Grade Group 5, while the remaining 4 of 24 cases (17%) were Grade Group 4 (3 cases Gleason score 4 + 4 = 8, 1 case Gleason score 3 + 5 = 8) (Figure 5, E). Tumor necrosis as a component of Gleason pattern 5 was observed in 21 of 24 cases (87.5%), with extensive necrosis observed in 5 of 24 cases (21%) (Figure 5, F).

The number of cores involved by high-grade desmoplastic foamy gland carcinoma ranged from 3 to 12 (Table 2), with a mean and median of 8 positive cores. The percentage of core involved, averaged per case, ranged from 30% to 90%, with a mean and median of 67% and 70%. Extraprostatic extension was observed in 6 of 21 needle biopsy cases (29%) (Figure 6, A and B). Seminal vesicle invasion was observed on needle biopsy in 3 of 21 cases (14%) (Figure 6, C and D).

The percentages of tumor volume involvement by high-grade desmoplastic foamy gland carcinoma in the 2 transurethral resection specimens were 50% and 60%.

In the prostate of the radical cystoprostatectomy specimen, there was extraprostatic extension and seminal vesicle invasion with a 0.5-mm metastasis to a pelvic lymph node (pT3b, pN1).

Aberrant staining for HMWCK in a nonbasal cell pattern (Figure 3, C and D; Figure 7, A through D) was observed in all cases (Table 2). In most cases there were only scattered cells and glands with aberrant HMWCK staining, yet in 6 of 24 cases (25%) extensive aberrant HMWCK staining of cancer cells was noted. No case had aberrant p63 expression.

PTEN loss (focal loss in 2 cases) was observed in 9 of 23 evaluable cases (39.1%), while 4 cases were interpreted as ambiguous (Table 3). Similarly, immunohistochemical staining of p53 as a surrogate marker for TP53 mutations was performed and showed that p53 nuclear accumulation was observed in 8 of 23 evaluable cases (34.8%), with 3 cases interpreted as ambiguous (Table 3).

Of the 18 cases where contributors provided a preliminary diagnosis, 5 diagnosed the case as Gleason scores 8-9 and 1 considered it as high-grade prostate cancer that was peculiar, low volume with skip areas, and challenging. An additional 2 contributors recognized the case as carcinoma, but undergraded the case as Gleason score 3 + 3 = 6 (Grade Group 1) and 4 + 3 = 7 (Grade Group 3), respectively. Other outside diagnoses were colon versus bladder adenocarcinoma (n = 1); intraductal prostate cancer (n = 1); Gleason pattern 4 versus intraductal prostate cancer (n = 1); prostate versus colorectal adenocarcinoma (n = 1); atypical glands, urothelial carcinoma versus degenerative prostate glands (n = 1); atypical glands (n = 1); atypical glands versus intraductal carcinoma (n = 2); urothelial carcinoma (n = 1); and Gleason score 8 versus benign prostate tissue (n = 1).

Cross talk between epithelial cells and the surrounding stromal components is fundamental in the context of normal prostate tissue for maintaining its homeostasis.^3,4  Tumor-associated stromal myofibroblasts and fibroblasts stimulated by transforming growth factor β1 combined with matrix-remodeling enzymes, lead to production of excess dense extracellular matrix, which can displace the normal prostate fibromuscular stroma.^4,5  Carcinoma-associated myofibroblasts and fibroblasts can promote cancer progression and invasive growth, while prostate cancer can promote a tumor-supporting environment by epigenetic or transcriptomic regulation of myofibroblasts and fibroblasts.^6–10  Thus, the activation of the host stromal microenvironment is a critical step in adenocarcinoma growth and progression; and desmoplasia, which is protumorigenic, is often associated with high-grade carcinoma or worse outcome.⁴ 

Host stromal reaction is highly variable in prostate cancer with most cases having little to no reactive stroma. A grading system to quantitate reactive stroma was initially proposed by the Baylor College group in a large patient cohort, using trichrome stain.¹¹  Tumors with 0% to 5% stroma were assigned a Reactive Stroma Grade (RSG) of 0. Tumors with 5% to 15% stroma were assigned RSG 1 and those with 15% to 50%, RSG 2. Tumors with greater than 50% reactive stroma were assigned RSG 3. Patients with RSG 3 had a significantly worse cancer-specific survival than those with RSG 1 or 2. The same authors subsequently converted their RSG system to a binary grading system with RSG 3 designated as stromogenic carcinoma, while RSG 1-2 was categorized as nonstromogenic. Patients with higher percentage of stromogenic carcinoma on radical prostatectomy have decreased biochemical-free and cancer-specific survival independent of Gleason grade.¹²  The prognostic role of reactive stromal pattern in Gleason score 6 (Grade Group 1) and Gleason score 7 (Grade Groups 2-3) has been validated in several independent cohorts.^13–17 

The first definitional morphologic feature of the prostate cancer subtype presented herein is the prominent desmoplastic stroma. The amount of stroma in the current series far exceeded the degree of stromal response typically seen in prostate adenocarcinoma. Overall, of the 16 patients with meaningful follow-up, 12 (75%) either had metastases or died due to prostate cancer, which follows with the above literature on stromogenic prostate cancers having a worse prognosis.

The second important morphologic feature of the prostate cancer subtype presented herein is the foamy gland morphology. Although the initial description of foamy gland carcinoma by Nelson and Epstein¹  in 1996 typically was Grade Group 1 with inconspicuous nucleoli, subsequently, high-grade foamy gland carcinoma was recognized. In 2001, Tran et al¹⁸  reported 6 cases of aggressive foamy gland carcinoma, with 5 of 6 cases Gleason score 7 or above with extraprostatic extension at radical prostatectomy. The first systematic study of high-grade foamy gland carcinoma of the prostate by Zhao and Epstein²  in 2009 analyzed a series of 55 cases, where the most common architectural pattern was cribriform (73%), followed by fused/poorly defined glands (55%), cords/single cells (11%), and solid sheets (5%). In 18 of the 55 cases (33%), there was at least a moderate stromal reaction. A moderate or greater stromal reaction was seen in 48% (11 of 23) of the cases with Gleason score 8 or above compared with 22% (7 of 32) of the cases with Gleason score 7. In 6 cases, the tumors were desmoplastic foamy gland carcinomas, as reported in the current study, 5 of which were Gleason scores 4 + 4 = 8 (Grade Group 4).

The third common feature of the desmoplastic foamy gland carcinoma seen in the current series is aberrant positive staining for HMWCK in a nonbasal cell pattern, which was observed in all cases, with extensive aberrant HMWCK staining in cancer cells seen in 6 cases (25%). Googe et al¹⁹  reported this phenomenon first in 1997, which is different from diffuse p63-expressing atrophic prostate cancer with strong p63 positive staining in a nonbasal pattern. The high frequency of aberrant HMWCK positivity in a nonbasal cell pattern in high-grade desmoplastic foamy gland carcinoma was in keeping with earlier findings showing that high-grade foamy gland cancer without desmoplasia had increased aberrant positive staining for HMWCK in a nonbasal cell pattern when compared with lower-grade foamy gland cancer and usual non–foamy gland cancer.² 

The final feature that typified high-grade desmoplastic foamy gland carcinoma was the frequent presence of intraductal carcinoma. In keeping with the aggressive clinical behavior of this prostate cancer subtype, the presence of intraductal carcinoma associated with usual prostate adenocarcinoma is associated with an adverse prognosis. Other findings seen in the current series associated with more aggressive prostate cancer is the loss of PTEN (39.1% of cases), and overexpression of P53 (at least 34.8% of cases). However, some rare TP53 gene truncation mutation has been reported to result in retained cytoplasmic p53 expression, such that the percentage of TP53 mutation in the current series could be higher.^20,21  PTEN loss has been associated with upgrading of prostate cancer from biopsy to radical prostatectomy and shorter time to prostate cancer recurrence after radical prostatectomy.^22–24  The frequency of PTEN loss increases with increasing Grade Group as well as pathologic stage.²⁴  TP53 tumor suppressor gene mutations have been identified as the most highly enriched non-AR alterations in castration-resistant prostate cancer and are associated with inferior outcomes to androgen-targeted therapies, and p53 nuclear accumulation is an early indicator of lethal prostate cancer.^20,21 

In summary, high-grade desmoplastic foamy gland carcinoma is rare and has distinct morphologic, immunohistochemical, and clinical features from the much more common typical foamy gland prostate adenocarcinoma. High-grade desmoplastic foamy gland adenocarcinoma is difficult to diagnose as cancer and to grade owing to (1) foamy gland cytoplasm with relatively bland nuclear features in certain areas; (2) prominent desmoplastic stromal reaction resulting in widely scattered partially obscured tumor; (3) frequently disrupted glands; and (4) nonspecific staining for HMWCK. Even when recognized as carcinoma, high-grade desmoplastic foamy gland adenocarcinoma differed significantly from usual prostate adenocarcinoma, raising concerns of a nonprostatic primary. Only one-third of the contributing pathologists were able to definitively diagnose and grade the tumor as Gleason score 8-10 prostate cancer with accuracy. It is important to recognize this subtype and its high-grade nature given its aggressive clinical course with frequent metastases and death.