Molecular Lymph Node Status for Prognostic Stratification of Prostate Cancer Patients Undergoing Radical Prostatectomy with Extended Pelvic Lymph Node Dissection Free

Prostate cancer is the most common cancer and the second to third leading cause of cancer death in American and European men (1–4). In patients treated with radical prostatectomy (RP) for localized prostate cancer evidence of lymph node (LN) metastasis poses a major prognostic factor of tumor recurrence as well as mortality (5, 6) and guides further treatment. In men with node-positive prostate cancer, early adjuvant androgen-deprivation therapy (ADT) improves cancer-specific survival (7). Moreover, adjuvant pelvic radiation therapy (RT) combined with ADT appears to have a beneficial impact on cancer-specific survival in men with limited pelvic LN involvement (8).

However, a substantial proportion of patients with node-negative prostate cancer suffer from tumor recurrence within a few years after surgery (9). This may result from metastases that remain undetected by standard examinations at the time of RP. Optimizing staging in prostate cancer patients undergoing RP is therefore of crucial interest.

Despite recent advances with novel imaging techniques, the gold standard for LN staging at RP is still a pelvic LN dissection (PLND), as all imaging techniques eventually miss small metastases (10–12). Herein, two factors are essential to correctly stage pelvic LNs. An extended PLND (ePLND) respecting the main lymphatic drainage sites of the prostate including the obturator, external, internal, and common iliac region is demanded in patients with elevated risk for LN metastasis (13, 14). Moreover, the detection rate of LN metastasis varies considerably according to the applied method for examination of the dissected lymphatic tissue. In comparison with standard histopathology and immunohistochemistry, molecular LN analysis using polymerase chain reaction (PCR) for the detection of cancer-enhanced transcripts shows the highest sensitivity for detection of LN metastases in prostate cancer and other solid tumors (14–20). We recently described a novel validated method for molecular LN analysis quantifying prostate-specific expression of Kallikrein 3 (KLK3) from fresh frozen LN specimens showing high sensitivity, specificity, and reproducibility (14). This approach might improve the identification of patients with high risk of tumor recurrence and guide adjuvant treatment. However, the prognostic role of molecularly detected small tumor cell deposits in LNs has not been clarified and is unknown in prostate cancer patients treated with RP and ePLND.

Consequently, we undertook a prospective trial to assess the prognostic impact of molecular LN analysis on biochemical recurrence in comparison with standard histopathology for detection of LN metastasis in prostate cancer patients undergoing RP with ePLND.

The present study was approved by the local ethics committee (ID 2607/09) and conducted in accordance with the Declaration of Helsinki. All patients signed review board–approved consent before participation. Between February 2010 and February 2013, LN specimens were obtained from 111 prospectively enrolled patients who underwent open RP and ePLND for intermediate or high-risk prostate cancer (Gleason score at biopsy ≥7 or PSA ≥10 ng/mL or clinical tumor extension ≥cT2b; ref. 21) at our institution. Exclusion criteria for study participation were presence of a second malignant tumor, previous pelvic radiation or ADT, acute or chronic infectious disease, and severe cardiopulmonary, renal, hepatic or hematopoietic disease.

RP and ePLND were performed according to a predefined template including bilateral, separate dissection of the obturator fossa, external iliac, internal iliac, and common iliac vessels (14). Distally, ePLND was limited by the femoral canal and proximally by the aortic bifurcation.

Perioperatively, LNs with a size ≥3 to 20 mm were bisected. One half and the lateral edge of the second half were formalin fixed and stained with hematoxylin and eosin (H&E) for histopathologic examination (pN0 or pN1). The remainder of the same LN was snap frozen within 30 minutes after removal and stored at −80°C for later RNA extraction. LNs with a size >20 mm were bisected, and resulting two pieces were examined like singular LNs <20 mm. LNs with a size <3 mm were only assessed by histopathology as tissue was insufficient for examination by both techniques. Standard histopathology comprised one section per 5 mm of LN tissue.

For molecular LN analysis, we used an analytically validated quantitative PCR (qPCR) assay established by our group (14). Briefly, LNs were homogenized and lysed for RNA extraction and complementary DNA (cDNA) was generated. We used commercially available Taqman probe and primer sets for detection of the prostate-specific antigen (PSA) transcript KLK3 and normalized qPCR results to endogenous reference gene expression of HPRT1 and UBC (Applied Biosystems) in relation to a calibrator sample [10 LNCaP cells spiked in 10⁶ peripheral blood mononuclear cells (PBMC) with a given relative gene expression of 1.0] using the ΔΔCT method. The sensitivity of the applied assay in serial dilutions was the detection of 1 LNCaP cell or 10 KLK3 copy numbers in 10⁶ PBMCs. The intra-assay and interassay coefficient of variation was <10% for all qPCR results, including LNCaP cell dilutions down to one single cell.

As control, 143 LNs from 25 male bladder cancer patients treated with cystoprostatectomy and histopathologic exclusion of prostate cancer were analyzed to determine a threshold for physiologic KLK3 expression. These LNs were assessed as negative by histopathology and qPCR for cytokeratin 20 expression in order to exclude bladder cancer metastasis.

We prospectively compared molecular analysis and histopathology to determine LN status in intermediate- and high-risk prostate cancer patients undergoing RP with ePLND. Previous results of this trial focusing on pelvic topography of LN metastases have been reported recently by our group (14). Here, we present follow-up data in an extended cohort assessing the association between LN status and biochemical recurrence. Results are reported in compliance with REMARK guidelines (22).

Follow-up information on the PSA level and further prostate cancer treatment was determined every 3 months in the first year following RP, every 6 months in the second year, and every 12 months thereafter. A confirmed postoperative PSA value of >0.2 ng/mL was considered as biochemical recurrence. Adjuvant radiotherapy and/or ADT after RP was administered based on histopathology results and defined as postoperative treatment without evidence of biochemical recurrence. All treating physicians were blinded to molecular LN results.

Data were analyzed using IBM SPSS Statistics version 24.0 and the statistical software package R (23). A threshold for molecular results (molN0 or molN1) was calculated identifying 99% of histopathologic true-negative LNs with a 99% level of confidence by using the R-package “tolerance” for estimating tolerance intervals based on a gamma distribution of KLK3 expression (24, 25). The Kruskal–Wallis test was applied to assess the association of LN status with clinical variables. The Spearman correlation coefficient rho was used to assess presence and magnitude of monotonous trends between the level of evidence of LN metastases (pN0/molN0<pN0/molN1<pN1/molN1) and ordinal clinical risk factors. Kaplan–Meier curves with log-rank statistics as well as univariable and multivariable Cox proportional hazard regression analyses were used to assess the association between LN status and bRFS. Hazard ratios (HR) and 95% confidence intervals (CI) were calculated. Robust variance estimators were used to account for the clustered nature of the data (multiple LNs per patient). All statistical tests were performed two-sided, and a P value of <0.05 was considered statistically significant.

Based on the control group we calculated a threshold of 1.47 for physiologic KLK3 expression in molecular negative (molN0) LNs versus pathologic KLK3 expression in molecular positive (molN1) LNs. Accordingly, patients were classified as negative for LN metastasis (pN0/molN0), positive exclusively by molecular examination (pN0/molN1), or positive by histopathologic and molecular examination (pN1/molN1; Supplementary Fig. S1).

Patient characteristics are given in Table 1. The mean age of patients was 66 years. Preoperatively, 29% of patients were diagnosed with intermediate and 71% of patients with high-risk prostate cancer.

In total, 3,173 LNs in 111 patients (median 27 LNs/patient; range, 9–78) were dissected, of which 2,411 LNs had a diameter ≥3 mm and were analyzed by both standard histopathology and qPCR (median 21 LNs/patient; range, 6–48; Fig. 1).

Of 111 patients, 28 (25%) had histopathologic and molecular positive LNs (pN1/molN1), 32 (29%) had histopathologic negative but molecular positive LNs (pN0/molN1), and 51 (46%) had histopathologic and molecular negative LNs (pN0/molN0). Thus, a total of 60 (54%) patients had node-positive disease by molecular analysis. Notably, no patient was positive by histopathologic but negative by molecular LN examination (pN1/molN0).

Of 2411 LNs analyzed by histopathology and molecular analysis, 65 (2.7%) were positive by histopathologic and molecular analysis, 3 (0.1%) were positive by histopathologic but negative by molecular analysis, 224 (9.3%) were negative by histopathologic but positive by molecular analysis, and the remaining 2119 (87.9%) LNs were negative by both methods. The three histopathologic positive but molecular negative LNs were present in patients with other molecular positive LNs and therefore there were no patients staged as pN1/molN0.

The number of LNs that were positive by molecular but negative by histopathologic analysis was more frequent in pN1/molN1 patients (n = 148, median 4; range, 0–21) than in pN0/molN1 patients (n = 76, median 2; range, 1–8).

A statistically significant association of LN status (pN0/molN0 vs. pN0/molN1 vs. pN1/molN1) was observed with preoperative risk group according to D´Amico classification, preoperative PSA level, prostate biopsy density, pathologic tumor extension, and Gleason score at RP (Table 2). These variables were most favorable in patients without LN metastases (pN0/molN0) and most unfavorable in patients with histopathologic and molecular LN metastases (pN1/molN1), while patients with exclusively molecular LN metastases (pN0/molN1) exhibited intermediate features. This correlation (pN0/molN0<pN0/molN1<pN1/molN1) was strongest for pathologic tumor extension (Spearman rho +0.53), followed by prostate biopsy density (Spearman rho +0.47), preoperative PSA level (Spearman rho +0.38), Gleason score at RP (Spearman rho +0.32), and D'Amico risk group (Spearman rho +0.23).

At a median follow-up of 48 months [interquartile range (IQR), 36–60] 52/111 (47%) patients developed biochemical recurrence, 11/51 (22%) patients with negative LNs (pN0/molN0), 21/32 (66%) patients harboring only molecular LN metastasis (pN0/molN1), and 20/28 (71%) patients with histopathologic and molecular LN metastasis (pN1/molN1). As shown in the Kaplan–Meier curves, a positive molecular LN examination despite negative histopathology (pN0/molN1) identified patients with an elevated risk for biochemical recurrence (Fig. 2A). Median biochemical recurrence-free survival (bRFS) was 9 months (95% CI, 0.0–20.1) in patients with histopathologic and molecular LN metastases (pN1/molN1), 24 months (95% CI, 1.7–46.3) in patients with only molecular LN metastases (pN0/molN1) and not reached in patients without LN metastases (pN0/molN0; log-rank test P < 0.001). In a multivariable Cox regression analysis, pN0/molN1 status [HR 4.1 (95% CI, 1.9–8.8)] and pN1/molN1 status [HR 6.3 (95% CI, 2.7–14.7)] showed a strong association with biochemical recurrence independent of clinically established risk factors including preoperative PSA level, postoperative tumor extension, Gleason score, and surgical margin status as well as adjuvant ADT or radiation therapy (Table 3A).

Next, we compared both LN sampling methods for prediction of biochemical recurrence. According to histopathology alone, median bRFS was 9 months (95% CI, 0.0–20.1) in 28 of 111 (25%) pN1-patients and was not reached in 83 of 111 (75%) pN0 patients (log-rank test P < 0.001; Fig. 2B). According to molecular LN analysis alone, median bRFS was 24 months (95% CI, 0.0–49.2) in 60/111 (54%) molN1 patients and was not reached in 51 of 111 (36%) molN0 patients (log-rank test P < 0.001; Fig. 2C). As displayed in their respective Kaplan–Meier curves, molecular LN status stratified patients better according to their risk of biochemical recurrence than histopathologic LN status. In a multivariable Cox regression analysis including both methods, molecular LN status [HR 4.1 (95% CI, 1.9–8.8), P < 0.001] but not histopathologic LN status [HR 1.5 (95% CI, 0.8–3.0), P = 0.198] was confirmed as an independent predictor of biochemical recurrence (Table 3B).

As adjuvant treatment may represent a bias here, we repeated this analysis omitting 7 patients with adjuvant treatment for pN1 disease from the model (ADT (n = 4), adjuvant radiation of the prostatic fossa including the pelvic lymphatic draining sites (n = 2) or both (n = 1; Supplementary Fig. S2). Consistently molecular but not histopathologic LN status was confirmed as an independent predictor of biochemical recurrence (P = 0.001 vs. P = 0.144; Supplementary Table S1A).

The superiority of molecular LN status for prediction of biochemical recurrence was confirmed when including the logarithmic maximum KLK3 expression in LNs as continuous variable in Cox regression models. Thus, molecular but not histopathologic LN status could be confirmed as independent prognostic factor of biochemical recurrence in the overall study cohort (P < 0.001 vs. P = 0.659; Table 3C) and the subgroup excluding patients with adjuvant treatment for pN1 disease (P < 0.001 vs. P = 0.401; Supplementary Table S1B).

In addition, we assessed the influence of extent of PLND on the detection rate of LN metastasis and the corresponding association with biochemical recurrence. Herein, analysis of LNs of the full ePLND template including obturator, external, internal, and common iliac region led to the highest detection rate of LN-positive patients translating into improved prediction of biochemical recurrence (Table 4). In contrast, analysis of more limited PLND templates (obturator vs. obturator + external iliac vs. obturator + external + internal iliac region) was associated with lower detection rates of LN metastasis and worse prognostic patient stratification.

Diagnostic and prognostic information was most enhanced when ePLND was combined with molecular LN analysis as compared with histopathology.

This is the first prospective trial comparing molecular LN examination in comparison with histopathology to detect LN metastasis in prostate cancer patients undergoing RP with ePLND with the aim to assess their association with biochemical recurrence. With a median of 27 dissected LNs per patient, a meticulous template ePLND was performed, which confidently surpassed the yield of at least 20 LNs for an adequate ePLND as postulated by Weingärtner and colleagues (26). In addition to 25% of patients with a positive histopathologic and molecular LN examination (pN1/molN1), molecular analysis identified 29% of patients harboring LN metastases despite negative histopathology (pN0/molN1). Both pN0/molN1 status and pN1/molN1 status were associated with shorter bRFS compared with patients with negative LNs (pN0/molN0) and predicted biochemical recurrence independent of established risk factors on multivariable analysis. Importantly, at comparison of both LN sampling methods on multivariable analysis, only molecular LN status (molN0 vs. molN1) but not histopathologic LN status (pN0 vs. pN1) remained an independent predictor of bRFS. The superiority of molecular LN analysis compared with histopathology was consistently observed in a multivariable analysis replacing categorical (molN0 vs. molN1) with continuous data of KLK3 expression in LNs. In addition, our results underline the necessity to perform a meticulous ePLND including obturator fossa, external, internal, and common iliac region in order to improve the detection of LN-positive patients and gain maximum prognostic information on risk of biochemical recurrence. Taken together, our results emphasize the risk of missing prognostically relevant pelvic LN metastases by standard histopathology and highlight the higher sensitivity of the qPCR-based molecular detection method in combination with ePLND translating into improved prognostic patient stratification. Thus, our data strongly support and extend two earlier prospective studies that reported molecular detection of metastases in histopathologic negative LNs as an independent prognostic marker of biochemical recurrence following RP or radiotherapy for localized prostate cancer (19, 20).

As a limitation of standard histopathology, small-volume metastases can be missed due to a sampling error on nodal slicing, or they may simply be missed on microscopic examination. A higher detection rate of LN metastases using histopathology can be achieved by step sectioning of LN tissue or application of immunohistochemistry. However, these methods, in addition to standard histopathology for LN examination, have been considered too time-consuming and expensive when compared against the diagnostic gain (27, 28). Moreover, in a retrospective analysis of prostate cancer patients treated with RP immunohistochemistry exhibited inferior sensitivity in comparison with molecular analysis for detection of LN metastasis (18).

As a main limitation of molecular LN analysis, morphologic information such as size of LN metastasis and capsular perforation cannot be determined. However, the qPCR-based method counteracts disadvantages of histopathology and provides higher detection rates by analyzing the LN specimen in toto. The method is based on previous tissue homogenization and subsequent automated observer-independent quantification of tumor cell–specific transcripts. Thus, it facilitates the capture of single tumor cells in one LN sample translating into early detection of lymphogenic tumor cell dissemination. This advantage is reflected in our data as molecular LN analysis detected pelvic metastatic disease at an earlier clinical stage than standard histopathology. At comparison with established clinicopathologic risk factors, patients without LN metastases had the most favorable, patients with only molecular positive LNs exhibited intermediate, and patients with histopathologic and molecular positive LNs showed the most unfavorable features.

Differences in three LNs staged as histopathologic positive but molecular negative might be due to the bisection of LNs. Of two LN halves, each half was examined only by one of both techniques. Therefore, small-volume metastases that were present only in one half of the LN might only have been detectable by one of both methods, an argument that also applies to LNs staged as histopathologic negative but molecular positive.

With emerging data emphasizing the importance of adjuvant treatment in LN-positive prostate cancer after RP, optimized staging methods for detection of LN metastasis will gain importance. In addition to a thorough ePLND, molecular LN analysis in patients treated with RP could become a powerful prognostic tool to guide further adjuvant treatment. Messing and colleagues were the first to report a benefit in progression-free, overall, and disease-specific survival in pN⁺ prostate cancer patients treated with adjuvant ADT after RP as compared with deferred ADT at clinical progression in a prospective, randomized trial (7). Certainly, the results of this trial are limited because of small sample size, with only 98 patients included, and because the trial was performed in the pre-PSA era. Nevertheless, recently Abdollah and colleagues reported in a contemporary cohort with over 1,100 patients a benefit in cancer-specific and overall survival of combined adjuvant ADT and pelvic radiotherapy for prostate cancer patients with limited nodal involvement after RP and ePLND (8). As the benefit from adjuvant treatment was clearly more apparent in patients with limited nodal involvement, the high sensitivity of molecular LN analysis with early detection of lymphogenic tumor cell dissemination might further improve patient stratification. A randomized controlled trial initiated by the German Associations of Radiologic Oncology and Urologic Oncology will further address the clinical benefit of adjuvant radiotherapy after RP targeting the pelvic lymphatic draining sites in patients with limited nodal involvement defined as histopathologic evidence of LN micrometastasis and up to two LN metastases (ART-2 study; EudraCT 2012-004322-24). Concordantly, a randomized-controlled trial including molecular LN analysis as a trigger of adjuvant pelvic RT could further elucidate the clinical utility of our approach.

An ePLND is recommended in European Urology guidelines for patients with elevated risk of LN involvement (13), and its diagnostic role is fully supported by this trial. In contrast, an ePLND is not a standard approach in the United States due to concerns about morbidity and lack of prospective data about improvement of outcome. If improved identification of node-positive disease by ePLND will actually lead to improved survival via selection of patients for adjuvant therapy in prospective trials, these data would further support implementation of an ePLND as a standard approach.

In prostate cancer patients treated with RP and ePLND molecular LN analysis provided a higher detection rate of clinically relevant LN metastases than standard histopathology identifying pN0 patients, who are at a high risk for biochemical recurrence. Moreover, molecular compared with histopathologic LN status stratified patients better according to their risk of biochemical recurrence. Thus, molecular analysis for the detection of LN metastasis improves the identification of patients who are at high risk for biochemical recurrence and might be applicable as a prognostic tool for guidance of adjuvant treatment decisions.

No potential conflicts of interest were disclosed.

Conception and design: M.M. Heck, M. Retz, R. Nawroth

Development of methodology: M.M. Heck, M. Retz, R. Nawroth

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.M. Heck, M. Bandur, M. Souchay, E. Vitzthum, G. Weirich, M. Autenrieth, H. Kübler, T. Maurer, M. Thalgott, K. Herkommer, J.E. Gschwend

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.M. Heck, T. Schuster

Writing, review, and/or revision of the manuscript: M.M. Heck, M. Retz, M. Bandur, M. Souchay, E. Vitzthum, G. Weirich, T. Schuster, M. Autenrieth, H. Kübler, T. Maurer, M. Thalgott, K. Herkommer, J.E. Gschwend, R. Nawroth

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.M. Heck, R. Nawroth

Study supervision: M.M. Heck, R. Nawroth

We thank Monika Moissl for her support in the laboratory with molecular LN analysis. This work was funded by the German Wilhelm-Sander Foundation (grant no. 2010.011.1; M.M. Heck, M. Retz, and R. Nawroth) and the Reinhard-Nagel Foundation (M.M. Heck) of the German Association of Urology.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.