Lymphovascular invasion is significantly associated with biochemical relapse after radical prostatectomy even in patients with pT2N0 negative resection margin.

Prostate Cancer and Prostatic Disease (2015) 18, 25–30 © 2015 Macmillan Publishers Limited All rights reserved 1365-7852/15 www.nature.com/pcan

ORIGINAL ARTICLE

Lymphovascular invasion is significantly associated with biochemical relapse after radical prostatectomy even in patients with pT2N0 negative resection margin K Mitsuzuka1,2, S Narita2,3, T Koie2,4, Y Kaiho1,2, N Tsuchiya2,3, T Yoneyama2,4, N Kakoi2,5, S Kawamura2,5, T Tochigi2,5, C Ohyama2,3, T Habuchi2,4 and Y Arai1,2 BACKGROUND: The significance of lymphovascular invasion (LVI) remains controversial, and the association of LVI with biochemical relapse was investigated in men treated with radical prostatectomy according to pathological results. METHODS: Data from 1268 patients undergoing radical prostatectomy between 2000 and 2009 were retrospectively reviewed. Clinicopathological variables were compared between LVI-negative and LVI-positive patients. Multivariate analyses by Cox proportional hazard model and Kaplan-Meier method were performed to identify risk factors for biochemical relapse in all patients, patients with pT2N0 and pT2N0 negative resection margin (RM). RESULTS: LVI information was available in 1160 cases, and LVI was seen in 121 cases (10.4%). Clinicopathological variables were significantly worse in LVI-positive patients than in LVI-negative patients. On multivariate analyses, PSA ⩾ 10 ng ml − 1, pathological Gleason score ⩾ 8, pathological T stage ⩾ 3, lymph node metastasis, positive RM and LVI were independent predictors for biochemical relapse in all patients. In patients with pT2N0, PSA ⩾ 10 ng ml − 1, pathological Gleason score ⩾ 8, positive RM and LVI were independent predictors for biochemical relapse. In patients with pT2N0 negative RM, LVI and pathological Gleason score ⩾ 8 were independent predictors for biochemical relapse (LVI; hazard ratio 3.809, 95% confidence interval 1.900–7.635, P-value o 0.001, Gleason score ⩾ 8; hazard ratio 2.189, 95% confidence interval 1.199–3.999, P-value = 0.011). With a median follow-up of 50 months, 5-year biochemical relapse-free survival in patients with pT2N0 negative RM was 95.7% in those with negative LVI in comparison to 85.3% in those with positive LVI (P o 0.001, log rank). CONCLUSIONS: LVI was consistently a significant predictor for biochemical relapse after radical prostatectomy in not only all patients but also in patients with pT2N0 and pT2N0 negative RM. These results strongly support the significance of LVI as a predictor for biochemical relapse. Prostate Cancer and Prostatic Disease (2015) 18, 25–30; doi:10.1038/pcan.2014.40; published online 21 October 2014

INTRODUCTION Lymphovascular invasion (LVI) has long been recognized as an essential step in metastases of various types of cancer.1–4 LVI could be a prognostic factor in upper urinary tract urothelial carcinoma,2 and LVI is independently associated with recurrence and survival in bladder cancer.1 However, the role of LVI in prostate cancer remains controversial. Recently, Ng et al.5 reviewed 19 studies examining the association between LVI in radical prostatectomy (RP) specimens and prostate cancer recurrence. In that review, 11 of the studies showed that LVI was an independent prognostic factor, while LVI was not associated with recurrence in the other 8 studies. The authors concluded that these previous studies had insufficient homogeneity to definitively establish LVI as an important independent prognostic factor. LVI in RP specimens might be associated with recurrence or progression of disease like urothelial carcinoma; however, the previous reports were likely to have small sample size, and other factors like extra prostatic extension, seminal vesicle invasion or positive resection margin (RM) may interfere to clarify the

significance of LVI on biochemical relapse. Therefore, we retrospectively analyzed 41100 cases of RP specimens and focused on patients with pT2N0 or pT2N0 negative RM to determine whether LVI was associated with biochemical relapse, because restriction to pT2N0 or pT2N0 with negative RM would remove influence of stage or margin status. MATERIALS AND METHODS This study was approved by the institutional review boards at each participating site, and institutional data-sharing agreements were established before initiation of the study. Between January 2000 and December 2009, 1268 consecutive patients with clinically localized prostate cancer underwent RP at Tohoku University, Hirosaki University, Akita University, and Miyagi Cancer Center. None of the patients had a prior history of prostate surgery nor did they undergo neoadjuvant hormonal or radiation therapy. RP was performed in an open, retropubic manner in all cases. Clinical information, including age, PSA, biopsy Gleason score and clinical T stage, was reviewed by urologists in each institution. The surgical specimen containing the prostate gland with seminal vesicles attached was serially sectioned perpendicular to the long axis of the gland.

1 Department of Urology, Tohoku University Graduate School of Medicine, Sendai, Japan; 2Michinoku Japan Urological Cancer Study Group, Sendai, Japan; 3Department of Urology, Akita University Graduate School of Medicine, Akita, Japan; 4Department of Urology, Hirosaki University Graduate School of Medicine, Hirosaki, Japan and 5Department of Urology, Miyagi Cancer Center, Natori, Japan. Correspondence: Dr K Mitsuzuka, Department of Urology, Tohoku University School of Medicine, 1-1 Seiryo-machi, Aoba-ku, Sendai 980-8574, Japan. E-mail: [email protected] Received 27 May 2014; revised 11 August 2014; accepted 27 August 2014; published online 21 October 2014

LVI in pT2N0 negative resection margin K Mitsuzuka et al

26 Pathological evaluation of surgical specimens, including pathological Gleason score (pGS), pathological T stage, RM, lymph node metastasis and tumor volume, was performed at each institution by a dedicated genitourinary pathologist according to the Gleason grading system and pathological stage based on the 2002 Tumor–Node–Metastasis classification. Total tumor volume was measured only at the Tohoku University and Hirosaki University using the computer-assisted image analysis (NIH Image, National Institutes of Health, Bethesda, MD, USA) and was defined as the sum of the volumes of individual cancer foci.6 LVI was defined as the unequivocal presence of tumor cells within endothelial-lined channels on routine light microscopic examination.7–10 Lymphatic or vascular invasion was not distinguished in this study. Patients were subgrouped according to pathological results (all/pT2N0/pT2N0 with negative RM) to analyze predictors for biochemical relapse and biochemical recurrence-free survival in each group. Of the 1268 cases of RP, LVI information was not available due to lack of description about LVI in 108 cases, which were excluded from this study. Of the 1160 cases with available LVI information, lymph node dissection was performed in 1013 cases (87.3%) with limited style, mainly in cases with intermediate- or highrisk disease. The follow-up schedule after RP involved a PSA assay every 3 months for the first 2 years, every 6 months for the following 3 years and then annually thereafter. The date of PSA recurrence was defined as the point at which the serum PSA level exceeded 0.2 ng ml − 1, when RP was carried out if the PSA did not decrease below 0.2 ng ml − 1 after surgery or when adjuvant therapy was initiated even if PSA did not exceed 0.2 ng ml − 1. Statistical analyses were performed with SAS 9.2 (SAS Institute, Cary, NC, USA). Clinicopathological parameters were analyzed using the Wilcoxon’s rank sum test or the Fisher’s exact test. To identify independent prognostic factors, the Cox proportional hazards regression model with variables, including age, PSA, biopsy Gleason score, clinical T stage, pGS, pathological T stage, lymph node metastasis, RM and LVI, was used. Tumor volume was not used as variables, because it was not measured in nearly half of the cases. Sixteen cases with unknown RM status were excluded from this model. Multivariate analyses were performed with all variables and selected variables by backward elimination. There were no clinically significant interactions between variables, which were confirmed by running proportional hazard models with two-factor interaction terms added to final model after backward elimination. No clinically significant deviation from proportional hazard assumption was also confirmed by running proportional hazards models with interaction terms of follow-up time and factors added to final model after backward elimination. The Kaplan–Meier method was used to determine biochemical relapse-free survival. A P-value o0.05 was considered significant.

Table 1. Characteristics of men treated with radical prostatectomy by the presence of lymphovascular invasion LVI-negative

LVI-positive

1039 (89.6%)

121 (10.4%)

Age o70 years ⩾ 70 years

705 (67.9%) 334 (32.1%)

95 (78.5%) 26 (21.5%)

PSA o10 ng ml − 1 ⩾ 10 ng ml − 1

752 (72.4%) 287 (27.6%)

73 (60.3%) 48 (39.7%)

Biopsy Gleason score ⩽6 7 ⩾8

358 (34.5%) 547 (52.6%) 134 (12.9%)

13 (10.7%) 66 (54.5%) 42 (34.7%)

Clinical T stage 1c 2 3

605 (58.2%) 384 (37.0%) 50 (4.8%)

59 (48.8%) 39 (32.2%) 23 (19.0%)

Pathological Gleason score ⩽6 7 ⩾8

152 (14.6%) 653 (62.8%) 234 (22.5%)

5 (4.1%) 50 (41.3%) 66 (54.5%)

Pathological T stage 2 3a 3b 4

750 233 47 9

46 46 25 4

Resection margin Negative Positive Unknown

762 (73.3%) 264 (25.4%) 13 (1.3%)

78 (64.5%) 43 (35.5%) 0 (0.0%)

1027 (98.8%) 12 (1.2%)

105 (86.8%) 16 (13.2%)

106 (18.5%) 316 (55.1%) 151 (26.4%)

5 (6.4%) 28 (35.9%) 45 (57.7%)

Number of patients

Lymph node metastasis Negative Positive Tumor volumea o0.5 ml 0.5–2.5 ml ⩾ 2.5 ml

P-value

0.017

0.006

o0.001

o0.001

o0.001

o0.001 (72.2%) (22.4%) (4.5%) (0.9%)

(38.0%) (38.0%) (20.7%) (3.3%) 0.021

o0.001

o0.001

RESULTS Frequency of LVI and clinicopathological variables according to the presence of LVI Of the 1268 patients treated with RP, LVI information was available in 1160 cases. Of these, LVI was seen in 121 cases (10.4%). The frequency of LVI was 20.6% in ⩾ pT3, 5.8% in pT2N0 and 5.8% in pT2N0 with negative RM. Table 1 shows the clinicopathological comparison according to the presence of LVI. Patients with LVI were likely to be younger, have higher PSA, biopsy Gleason score, clinical T stage, pathological T stage, pGS and tumor volume and have a higher rate of positive RM and lymph node metastasis.

relapse than pGS ⩾ 8 in patients with pT2N0 negative RM (LVI: hazard ratio 3.809, 95% confidence interval 1.900–7.635, Po 0.001/pGS ⩾ 8: hazard ratio 2.189, 95% confidence interval 1.199–3.999, P = 0.011).

Predictors for biochemical relapse according to various pathological states Multivariate analyses by Cox proportional hazard model were performed to predict factors for biochemical relapse in all cases, pT2N0 cases and pT2N0 negative RM cases. In all patients, PSA ⩾ 10 ng ml − 1, pGS ⩾ 8, pT ⩾ 3, lymph node metastasis, positive RM and LVI were significant predictors for biochemical relapse (Table 2). In patients with pT2N0, PSA ⩾ 10 ng ml − 1, pGS ⩾ 8, positive RM and LVI were significant predictors for biochemical relapse (Table 3). In patients with pT2N0 negative RM, pGS ⩾ 8 and LVI were significant predictors for biochemical relapse (Table 4). LVI was consistently a significant factor despite the pathological results, and LVI was a more significant predictor for biochemical

Biochemical relapse-free survival according to various pathological states With a median follow-up of 50 (1–131) months, 5-year biochemical recurrence-free survivals for LVI-negative and LVI-positive patients were 89.1% and 65.5%, respectively, in all patients (P o 0.001, log rank), 94.5% and 86.7%, respectively, in patients with pT2N0 (P = 0.003, log rank) and 95.7% and 85.3%, respectively, in patients with pT2N0 negative RM (P o0.001, log rank) (Figure 1). Figure 2 shows biochemical relapse-free survival in patients with pT2N0 negative RM according to pGS and LVI states. Five-year biochemical recurrence-free survival in pT2N0 negative RM with pGSo 8 and LVI-negative was 96.7%, which decreased significantly to 90.1% in pT2N0 negative RM with pGS ⩾ 8 and

Prostate Cancer and Prostatic Disease (2015), 25 – 30

Abbreviation: LVI, lymphovascular invasion aTumor volume was measured only in Tohoku and Hirosaki Universities (651 cases).

© 2015 Macmillan Publishers Limited


790 354 818 326 970 174 1077 67 849 295 786 358 1117 27 839 305 1024 120

(69.1%) (30.9%) (71.5%) (28.5%) (84.8%) (15.2%) (94.1%) (5.9%) (74.2%) (25.8%) (68.7%) (31.3%) (97.6%) (2.4%) (73.3%) (26.7%) (89.5%) (10.5%)

147 67 106 108 157 57 187 27 101 113 74 140 191 23 97 117 156 58

(18.6%) (18.9%) (13.0%) (33.1%) (16.2%) (32.8%) (17.4%) (40.3%) (11.9%) (38.3%) (9.4%) (39.1%) (17.1%) (85.2%) (11.6%) (38.4%) (15.2%) (48.3%)

Number of BCR (%) Reference 1.016 Reference 2.938 Reference 2.260 Reference 2.639 Reference 3.998 Reference 4.993 Reference 13.975 Reference 4.085 Reference 3.816

HR — (0.761, 1.356) — (2.247, 3.841) — (1.669, 3.062) — (1.762, 3.953) — (3.055, 5.234) — (3.765, 6.621) — (8.975, 21.762) — (3.118, 5.351) — (2.821, 5.162)

(95% CI) — 0.915 — o0.001*** — o0.001*** — o0.001*** — o0.001*** — o0.001*** — o0.001*** — o0.001*** — o0.001***

P-value Reference 0.838 Reference 1.681 Reference 1.047 Reference 0.916 Reference 2.253 Reference 2.179 Reference 3.383 Reference 2.420 Reference 2.069

HR — (0.619, 1.135) — (1.261, 2.241) — (0.736, 1.489) — (0.581, 1.445) — (1.645, 3.087) — (1.567, 3.030) — (2.044, 5.599) — (1.793, 3.268) — (1.483, 2.887)

(95% CI)

P-value — 0.253 — o 0.001*** — 0.799 — 0.706 — o 0.001*** — o 0.001*** — o 0.001*** — o 0.001*** — o 0.001***

Multivariate-all variables

Reference 2.247 Reference 2.174 Reference 3.196 Reference 2.374 Reference 2.126

Reference 1.668

HR

— (1.680, 3.006) — (1.563, 3.022) — (1.987, 5.141) — (1.762, 3.198) — (1.545, 2.927)

— (1.252, 2.222)

(95% CI)

— o0.001*** — o0.001*** — o0.001*** — o0.001*** — o0.001***

— o0.001***

P-value

Multivariate-backward elimination

549 235 616 168 691 93 758 26 637 147 670 114 740 44

(70.0%) (30.0%) (78.6%) (21.4%) (88.1%) (11.9%) (96.7%) (3.3%) (81.3%) (18.7%) (85.5%) (14.5%) (94.4%) (5.6%)

46 26 45 27 61 11 70 2 48 24 50 22 61 11

(8.4%) (11.1%) (7.3%) (16.1%) (8.8%) (11.8%) (9.2%) (7.7%) (7.5%) (16.3%) (7.5%) (19.3%) (8.2%) (25.0%)

Number of BCR (%) Reference 1.313 Reference 2.309 Reference 1.330 Reference 0.730 Reference 2.488 Reference 2.923 Reference 2.998

HR — (0.811, 2.124) — (1.433, 3.722) — (0.698, 2.533) — (0.178, 2.989) — (1.522, 4.067) — (1.769, 4.828) — (1.574, 5.709)

(95% CI) — 0.268 — o0.001*** — 0.386 — 0.661 — o0.001*** — o0.001*** — o0.001***

P-value

Reference 1.238 Reference 2.043 Reference 0.924 Reference 0.653 Reference 2.155 Reference 2.654 Reference 2.647

HR

— (0.758, 2.022) — (1.256, 3.324) — (0.445, 1.917) — (0.158, 2.703) — (1.228, 3.782) — (1.595, 4.418) — (1.368, 5.121)

(95% CI)

P-value — 0.394 — 0.004** — 0.832 — 0.557 — 0.007** — o0.001 *** — 0.004 **


Reference 2.141 Reference 2.686 Reference 2.594

Reference 2.051

HR

— (1.299, 3.530) — (1.620, 4.454) — (1.347, 4.993)

— (1.267, 3.320)

(95% CI)

— 0.003** — o0.001*** — 0.004**

— 0.003**

P-value


Abbreviations: BCR, biochemical relapse; bGS, biopsy Gleason score; CI, confidence interval; cT, clinical T stage; HR, hazard ratio; LVI, lymphovascular invasion; pGS, pathological Gleason score; RM, resection margin. *Po 0.05; **Po0.01; ***Po 0.001.

LVI

RM

pGS ⩾ 8

cT ⩾ 3

bGS ⩾ 8

PSA ⩾ 10

No Yes No Yes No Yes No Yes No Yes No Yes No Yes

Number of patients (%)

Univariate

Cox proportional hazard analyses to identify predictors of biochemical relapse after radical prostatectomy in pT2N0 cases

Age ⩾ 70 years

Variables

Table 3.

Abbreviations: BCR, biochemical relapse; bGS, biopsy Gleason score; CI, confidence interval; cT, clinical T stage; HR, hazard ratio; LVI, lymphovascular invasion; pGS, pathological Gleason score; pN, lymph node metastasis; pT, pathological T stage; RM, resection margin. *Po 0.05; **Po0.01; ***Po0.001.

LVI

RM

pN

pT ⩾ 3

pGS ⩾ 8

cT ⩾ 3

bGS ⩾ 8

PSA ⩾ 10

No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes


Univariate

Cox proportional hazard analyses to identify predictors of biochemical relapse after radical prostatectomy in all cases

Age ⩾ 70 years

Variables

Table 2.


27



467 203 540 130 585 85 646 24 548 122 631 39

(69.7%) (30.3%) (80.6%) (19.4%) (87.3%) (12.7%) (96.4%) (3.6%) (81.8%) (18.2%) (94.2%) (5.8%)

32 18 38 12 40 10 48 2 33 17 39 11

(6.9%) (8.9%) (7.0%) (9.2%) (6.8%) (11.8%) (7.4%) (8.3%) (6.0%) (13.9%) (6.2%) (28.2%)

Number of BCR (%) Reference 1.305 Reference 1.325 Reference 1.677 Reference 0.957 Reference 2.645 Reference 4.589

HR

HR Reference 1.326 Reference 1.305 Reference 0.975 Reference 0.940 Reference 2.068 Reference 4.096

P-value — 0.367 — 0.396 — 0.147 — 0.952 — 0.001** — o0.001***

(95% CI) — (0.732, 2.327) — (0.692, 2.536) — (0.834, 3.372) — (0.230, 3.979) — (1.471, 4.756) — (2.339, 9.005)

— (0.732, 2.404) — (0.671, 2.537) — (0.435, 2.185) — (0.222, 3.978) — (1.037, 4.125) — (2.019, 8.310)

(95% CI) — 0.352 — 0.432 — 0.951 — 0.933 — 0.039* — o0.001***

P-value


P-value

— 0.011* — o0.001***

(95% CI)

— (1.199, 3.999) — (1.900, 7.635)

HR

Reference 2.189 Reference 3.809


Biochemical relapse-free survival



Abbreviations: BCR, biochemical relapse; bGS, biopsy Gleason score; CI, confidence interval; cT, clinical T stage; HR, hazard ratio; LVI, lymphovascular invasion; pGS, pathological Gleason score; RM, resection margin. *Po0.05; **Po 0.01; ***Po0.001.

LVI

pGS ⩾ 8

cT ⩾ 3

bGS ⩾ 8

PSA ⩾ 10

No Yes No Yes No Yes No Yes No Yes No Yes


Univariate

Cox proportional hazard analyses to identify predictors of biochemical relapse after radical prostatectomy in pT2N0 cases with negative RM

Age ⩾ 70 years

Variables

Table 4.


28

LVI (-) 1039 884

Months 646 414 149

LVI (+) 121 113 87 51 18

Months

LVI (-) 750 628 462 294 100

LVI (+) 44 41 34 19 7

Months

LVI (-) 631 529 388 244 78

LVI (+)

39

37

30

17

7

Figure 1. Biochemical relapse-free survival by pathological results. LVI, lymphovascular invasion; RM, resection margin.

LVI-negative, and further decreased to 82.5% in pT2N0 negative RM with pGS ⩾ 8 and LVI-positive.

DISCUSSION The results of the present study showed that LVI could be an independent predictor for biochemical relapse in patients treated following RP. The frequency of LVI was 10.4% in all patients, which



1) pGS

Serum luteinizing hormone concentration is significantly associated with recovery of urinary function after radical prostatectomy.

ERG expression is associated with increased risk of biochemical relapse following radical prostatectomy in early onset prostate cancer.

Robot-assisted radical prostatectomy significantly reduced biochemical recurrence compared to retro pubic radical prostatectomy.

Abdominoperineal excision following preoperative radiotherapy for rectal cancer: unfavorable prognosis even with negative circumferential resection margin.

Positive resection margins may not reflect the true margin in patients undergoing radical prostatectomy.

CT in patients with biochemical relapse after radical prostatectomy: a prospective study in 28 patients.

Editorial comment to high percent tumor volume predicts biochemical recurrence after radical prostatectomy in pathological stage T3a prostate cancer with a negative surgical margin.

Stage II-IV lung cancer cases with lymphovascular invasion relapse within 2 years after surgery.

Prediction of outcome following early salvage radiotherapy among patients with biochemical recurrence after radical prostatectomy.

Location of positive surgical margin and its association with biochemical recurrence rate do not differ significantly in four different types of radical prostatectomy.

Impact of adjuvant chemotherapy in patients with upper tract urothelial carcinoma and lymphovascular invasion after radical nephroureterectomy.

Impact of lymphovascular invasion on recurrence and progression rates in patients with pT1 urothelial carcinoma of bladder after transurethral resection.

Prognostic factors for survival after curative resection of distal cholangiocarcinoma: perineural invasion and lymphovascular invasion.

Salvage radiotherapy in patients with prostate cancer and biochemical relapse after radical prostatectomy : Long-term follow-up of a single-center survey.

BMI is associated with larger index tumors and worse outcome after radical prostatectomy.

Is a wider angle of the membranous urethra associated with incontinence after radical prostatectomy?

Salvage radiotherapy after radical prostatectomy: prediction of biochemical outcomes.

Statin use after radical prostatectomy reduces biochemical recurrence in men with prostate cancer.

Effect of Sulforaphane in Men with Biochemical Recurrence after Radical Prostatectomy.

Expression of cyclooxygenase-2 in cervical cancer is associated with lymphovascular invasion.

Predictive factors for biochemical recurrence in radical prostatectomy patients.

A risk-adjusted definition of biochemical recurrence after radical prostatectomy.

The distance of proximal resection margin dose not significantly influence on the prognosis of gastric cancer patients after curative resection.

Prognostic significance of lymphovascular invasion in node-negative gastric cancer.