Frequency of Positive Surgical Margin at Prostatectomy and Its Effect on Patient Outcome

A positive surgical margin at prostatectomy is defined as tumor cells touching the inked edge of the specimen. This finding is reported in 8.8% to 42% of cases (median about 20%) in various studies. It is one of the main determinants of eventual biochemical (PSA) failure, generally associated with a doubled or tripled risk of failure. The effect of a positive margin on outcome can be modified by stage or grade and the length, number and location of positive margins, as well as by technical operative approach and duration of operator experience. This paper tabulates data from the past decade of studies on margin status.


Definition of a Positive Surgical Margin (PSM) in Radical
Prostatectomy Specimens. As with all surgical specimens resected for cancer, the margins of a prostatectomy specimen are inked, usually using one color dye for the right side and one for the left. It is the pathologist's task to assess the microscopic slides and determine the proximity of tumor glands or cells to the ink to decide whether there is a definite positive surgical margin (PSM) (Figure 1).
A fundamental question is whether a tumor focus that is close to, but not touching, the resection margin ( Figure 2) holds the same implications as a PSM. This question was first answered by Epstein and Sauvageot in 1997, in a study of 101 cases [1]. They found that patients with biochemical progression were no more likely to have tumor close to the margin than those without progression. Emerson et al., confining their study to just 278 margin-negative wholemount prostate cases, validated that the closest distance between tumor and resection margin was not a significant predictor of PSA recurrence by univariate or multivariate analysis [2]. Thus, it was the consensus of the International Society of Urological Pathology in 2009 not to mention in written reports if tumor merely approaches but does not touch the margin [3]. This contrasts with the practice in other types of specimens such as breast lumpectomy specimens, in which the distance of tumor close to the margin is reported and does matter for outcome.
A PSM is a strong determinant of the probability of biochemical failure and is at least as important as grade, stage, and preoperative serum prostate-specific antigen (PSA). In unselected contemporary studies the PSM rate ranges from 8.8% [4] to 37% [5]. The interobserver reproducibility of designation of a PSM by urologic pathologists, using the definition of tumor on ink, has been shown to be good to excellent. The kappa value is 0.73 for definitive surgical margin status [6]. This supports the validity of many studies in concluding that, compared to negative surgical margin (NSM) status, a PSM correlates with a significant rise in biochemical failure rate. The purpose of this paper is to provide a compendium for urologists and their patients of all that is known about prostate margin status as an outcome predictor.

Methods
A review of papers pertaining to prostate margin status and its effect on outcome was undertaken using PubMed searches from 1997 to the present.

Can Prostate Biopsy Results Predict Margin Status?
We undertook a study a few years ago to determine the extent to which prostate biopsy results could predict cancer at prostatectomy that is unifocal, unilateral, margin-negative, and of small volume [20]. These four factors are the main criteria for choosing minimally invasive therapies such as targeted focal ablation of the prostate, as alternatives to radical prostatectomy. Unilateral cancer at prostatectomy was predicted by unilateral cancer in the biopsy (OR, 4.30) and unifocal cancer in the biopsy (OR, 2.63). In that study, negative surgical margins were predicted by unilateral cancer in the biopsy (OR 2.53, positive predictive value 82%). Therefore, biopsy findings can strongly predict prostatectomy margin status and other findings. (Table 1). In the past decade, nonrobotic or robotic laparoscopic techniques have been increasingly used in place of conventional open radical prostatectomy. The laparoscopic approaches are often considered superior for continence and potency [8,11,12,14,16]. Most studies involving prostate   [18].

Comparison of PSM Rates by Technical Approach
PSM rate for robotic approaches was found to be significantly worse than that for open ones (P = .007) in one study [10]; however, two other studies found open approaches superior to the robotic ones [17,18]. In the study that found the open approach better, the result was confounded by nerve sparing, so robotic prostatectomies showed a nonsignificant trend toward lower PSM for a non-nerve-sparing approach (P = .09) [10]. When the anterograde open approach was compared with the retrograde approach, significantly fewer PSMs were found by retrograde approach (P = .03) [9].
In a comparison of robotic versus nonrobotic laparoscopic approaches, one study found the robotic method superior [8]. Another found that the outcome was highly stage dependent, with 7% of pT2 patients with biochemical failure as opposed to 34% of pT3 patients [11]. Failure could also depend on number of positive margins [16]. In a study evaluating the robotic approach, a lower PSM rate was achieved by cold incision of the dorsal venous complex before suture ligation [12]. (Table 2). In the above comparison of surgical approaches, it must be noted that the new laparoscopic approaches have a demonstrable learning curve. That is, in three studies conducted in the middle of the 2000-2010 decade, the PSM rate improved after a few years of practice [21][22][23]. While a significant decrease in PSM rate occurred over time with a laparoscopic approach, PSM held steady for open procedures during the same time period [19]. Even  with the open approach, during the 1990s and early 2000s, one study had noted that there was also a learning curve with respect to the PSM rate [24].

Comparison of PSM Rates by Duration of Surgical Experience
It is a bit disconcerting but it also must be admitted that individual surgeons may vary in their frequency of PSMs. In a study of 4,629 men operated on by open prostatectomy by one of 44 surgeons, for the 26 surgeons who each treated >10 patients, the rate of PSM ranged from 10% to 48% [33]. A 6-fold difference was even reported for the same surgeon at different institutions [13]. (Table 3). PSM rates in studies not comparing approaches ranged from 13% [25] to 42% [31] with a median 23% [27].

Tumor Stage (
The HR for failure with a PSM seems to increase with increasing Gleason score [4,35,42,44]. In one study [34], however, after controlling for Gleason score, a PSM versus NSM with Gleason ≤7 was significantly predictive of failure, while PSM versus NSM with Gleason ≥8 was not (P = .115). Finally, Cao et al. noted that the Gleason score at the positive margin was predictive of biochemical recurrence [59]. Also, as the Gleason score of the main tumor rose, the concordance with the grade at the margin diminished: 99% for score 6 but 38% for score 9. By multivariate analysis, Gleason score at the margin predicted biochemical failure (P < .05) [59]. (Table 6). Three studies addressed the prostate cancer-specific death rate in the presence of a PSM. Two studies, one based on the SEER cancer data registry [45], found a significantly higher death rate at 10 years in the presence of a PSM [34,45], namely, 0.86% versus 0.33% (P < .001) and 2.6% versus 0.6% which was significant (P = .006). In another study, from the Mayo Clinic registry, a PSM was not a significant predictor of death among 11,729 cases (P = .15), but did predict death in the subset that was stage pT3 [34]. (Table 7) and by Location of PSM (Table 8). Many pathologists report the length of a PSM. Using categorical PSM length cut-offs between 3 mm and 10 mm, length significantly affected outcome in many [36,41,[47][48][49]58] but not all [50][51][52] studies. Emerson et al. [53] found a PSM length >3 mm to be a significant outcome predictor by univariate analysis but it fell short of significance by multivariate analysis (P = .076) [53]. Moreover, the length of PSM by frozen section predicted residual tumor in additionally resected neurovascular bundles by multivariate analysis (P < .001) [55].

PSA Failure Rates after a PSM Are Influenced by Length and Number of PSM
The number of PSMs probably lacks predictive value. In most studies, number of PSM was not significant for     [29,31,47,49]. In two studies, multiple PSMs as opposed to a single PSM predicted failure (HR 1.4, P = .002 by multivariate analysis or HR = 2.19) [54,58]. In another study, number of PSMs carried only borderline significance when ≥3 foci were positive compared to one (P = .06) and not significant for 2 foci compared to one [50]. Emerson et al. found that PSM number predicted failure by univariate analysis (P = .037) but lost most of its predictive value when adjusted for Gleason score (P = .076) [53]. The most common location of a PSM was in the posterior or posterolateral prostate [41,47,49], although one study found PSM equally common at the apex [24]. A positive apical soft tissue margin appears more consequential than a prostatic tissue margin [56]. Eastham et al. noted that the elevated risk of a posterior PSM means that "efforts to maintain adequate tissue covering including the routine excision of Denonvilliers' fascia and a component of the fat of the anterior rectal wall should be made in all patients. . ." [24]. Broken down by various sites, a posterolateral PSM predicted failure in most studies [24,48] but not all [49].
Comparing various sites of PSM, the effect of an apical PSM was not significantly different from PSM at posterolateral or other sites [29,52,58], and another study concluded that the PSM location seemed not to predict failure [53]. However, in two studies, a positive posterolateral margin predicted failure while the apical margin did not [24,57]. Possibly, residual apical tumor is less viable than residual tumor in the posterolateral region.

Conclusion
Prostate margin status is an important determinant of patient outcome after radical prostatectomy. In a 2010 College of American Pathologists survey, this feature was missing from 1% of pathology reports [60], thus the inclusion of this and other essential features is a quality assurance concern