- Top of page
- Materials and methods
Objective: The role of the lymph node dissection (LND) in conjunction with nephroureterectomy (NU) in upper tract urothelial cell carcinoma (UT-UCC) remains undefined. We evaluated the manner in which the LND was applied at NU, the patterns of lymph node (LN) involvement and the preoperative variables that could identify patients at high risk for lymph node metastasis (LNM).
Methods: We examined clinical, radiological and pathological records of patients who underwent NU for UT-UCC between 1985 and 2004. The central pathology laboratory reviewed all specimens and graded tumors using the 2002 World Health Organization/International Society of Urologic Pathologists grading system.
Results: Of the NU performed in 252 patients for UT-UCC, 105 (42%) were N0, 28 (11%) N+ and 119 (47%) Nx. Some form of LN resection was performed with NU in 53% of patients, with a median of four LN sampled (interquartile range, 2–10). After adjustment for tumor and patient characteristics, surgeon remained a significant predictor of LN resection (P < 0.0005). Of the evaluated variables, suspicious LN on preoperative computed tomography, present in 60% of N+ patients, was the only preoperative variable associated with the pathological finding of LNM (P < 0.0005).
Conclusions: LND in patients with UT-UCC is surgeon-dependent. Given the prognostic importance of LN status and the limited accuracy of preoperative staging of the regional LN, surgeons should perform a regional LND at the time of NU.
- Top of page
- Materials and methods
Despite encouraging reports of renal-sparing treatment of low-grade ureteral or renal pelvis tumors in selected patients,1,2 nephroureterectomy (NU) with bladder cuff resection remains the gold standard of treatment for high-grade upper tract urothelial cell carcinoma (UT-UCC).1 The natural history of the disease, specifically its multifocal nature, frequent recurrences after conservative therapy, propensity for local invasion and poor outcome once invasive disease is present, strongly argues for aggressive treatment for most cases. As with bladder cancers, high-grade invasive UT-UCC is frequently associated with regional lymph node metastasis (LNM), which can precede the development of distant disease. Extrapolating results from urothelial bladder cancer studies,3 it seems reasonable to believe that lymph node dissection (LND) in conjunction with NU may provide not only useful staging and prognostic information but also a therapeutic benefit in selected patients with UT-UCC. Nonetheless, it is impossible to draw any consistent conclusion on either way as previously published NU series present few details regarding practice patterns, including selection criteria, indications or LND templates. Published practice patterns vary greatly among institutions with patients receiving LND from large NU series ranging from 44%4 to 67%5 for open NU series;1,26–12 therefore, the role of the regional retroperitoneal LND in patients with UT-UCC remains undefined.
Our objective was to describe the manner in which the LND was applied at NU and the pattern of LNM at NU in patients with primary UT-UCC. We also sought to determine preoperative predictors of LNM to identify a population of patients who may eventually benefit from LND and/or could be selected for neoadjuvant therapy trials. Additionally, we analyzed the pathological features associated with LNM to identify which may be used as high-risk characteristics on preoperative biopsy or endoscopic resection specimens.
Materials and methods
- Top of page
- Materials and methods
Clinical, pathological and radiographic information was collected in 255 NU performed in 252 patients between December 1985 and March 2004 at Memorial Sloan-Kettering Cancer Center. Three patients underwent NU of the contralateral unit; data for only the first NU were included in the analysis. All NU were performed either open (n = 248) or laparoscopically (n = 7) by one of 10 specialized urological oncologist. The use, extent and template of the LND was at the surgeon’s discretion.
After internal review board approval, preoperative information was recorded for each patient, including age, gender, body mass index (BMI), prior bladder cancer, prior radical cystoprostatectomy, tobacco history (former or current vs non-smoker), grade of the biopsy (negative, low grade or high grade), tumor side, urine cytology result (positive, suspicious or negative), presence of suspicious regional lymph node (LN) enlargement on preoperative computed tomography (CT), administration of neoadjuvant therapy, and age-adjusted Charlson comorbidity index.13 All operative notes were reviewed in detail to determine intraoperative LN status, if described, as well as the extent and location of the LN dissection or sampling.
All specimens, including location and number of the dissected and metastatic LN, were analyzed per protocol at the pathology department of Memorial Sloan-Kettering Cancer Center (MSKCC). Tumors were graded using the World Health Organization (WHO) grading system (low, intermediate, and high grade), and staged according to the 2002 American Joint Committee on Cancer and tumor-node-metastasis (AJCC-TNM) staging system. Tumors were evaluated for location (renal pelvis, ureter or both), size, pathological stage, laterality, parenchymal invasion, capsular invasion, sinus invasion, surgical margins status (defined as malignant cells at the inked margin), ureteral margin status (presence of carcinoma in situ and/or invasive carcinoma at the ureteral line of resection), lymphovascular invasion, cell differentiation (presence of squamous, glandular or sarcomatoid features), associated carcinoma in situ, multifocality and perineural invasion.
Patients were followed postoperatively with physical examination, periodic cystoscopy, urine cytology, serum evaluations and imaging studies at the discretion of their individual physicians. Cancer-specific mortality was defined as death in the setting of local recurrence or distant metastasis as recorded by death certificates and/or medical record review. Cancer-specific survival was defined as the time from the NU to the date of last follow up or cancer-specific death; patients with non-disease related deaths were censored at the time of death.
Administration of perioperative chemotherapy varied from case to case over time. Thirty-one selected patients received perioperative chemotherapy, 17 received adjuvant therapy (most of them methotrexate, vinblastine, adriamycin, cisplatinum combination [MVAC] and platinum-based regimens), another seven received neoadjuvant (mostly MVAC and platinum-based regimens) and the other seven were unclear whether they received adjuvant or neoadjuvant. Among all N+-patients, 15 received some form of perioperative chemotherapy: nine adjuvant, two neoadjuvant and four were unclear.
Logistic or linear regression included the aforementioned preoperative variables to identify factors associated with LND and LNM at the time of NU. When analyzing predictors of LND and number of LN retrieved, we fitted a multivariate mixed model (using the xtlogit command in Stata, College Station, TX, USA) in which surgeon was entered as a random effect. In the mixed model, surgeons with less than 20 cases were grouped together in the same cluster. BMI was analyzed as a continuous variable. Neoadjuvant therapy was not entered into the predictive model as only nine patients received preoperative chemotherapy. Histological grade and presence of perineural invasion could not be used in the model because, respectively, only one-third of the patients were biopsied and there were no cases of perineural invasion among patients without LNM (Table 1). When analyzing predictors of LNM, we excluded patients who did not have LND performed (Nx patients) because they provided no information about LNM. Due to the low number of patients with LNM (n = 28), associations with LNM were assessed univariately. Statistical analyses were conducted on Stata 8.2.
Table 1. Patient characteristics according to lymph node status
| ||Nx n (%)||N0 n (%)||N1-3 n (%)||Total n (%)|
|Number of patients||119 (47%)||105 (42%)||28 (11%)||252|
|Age median (IQR) (n = 252)|| 70 (61, 76)|| 70 (62, 75)||68 (56, 76)|| 69 (60, 75)|
|Gender (n = 252)|
| Female|| 33 (28%)|| 41 (39%)||12 (43%)|| 86 (34%)|
| Male|| 86 (72%)|| 64 (61%)||16 (57%)||166 (66%)|
|Side (n = 252)|
| Right|| 58 (49%)|| 56 (53%)||18 (64%)||132 (52%)|
| Left|| 61 (51%)|| 49 (47%)||10 (36%)||120 (48%)|
|Preoperative urine cytology (n = 216)|
| Negative|| 43 (42%)|| 24 (26%)|| 3 (14%)|| 70 (32%)|
| Positive|| 60 (58%)|| 68 (74%)||18 (86%)||146 (68%)|
|Biopsy (n = 93)|
| Negative or indeterminate|| 12 (33%)|| 7 (15%)|| 1 (9%)|| 20 (22%)|
| Low Grade|| 21 (58%)|| 25 (54%)|| 6 (55%)|| 52 (56%)|
| High Grade|| 3 (8%)|| 14 (30%)|| 4 (36%)|| 21 (23%)|
|Ever smoked (n = 208)|
| No|| 24 (25%)|| 21 (23%)|| 5 (24%)|| 50 (24%)|
| Yes|| 73 (75%)|| 69 (77%)||16 (76%)||158 (76%)|
|Prior bladder cancer (n = 247)|
| No|| 60 (52%)|| 57 (55%)||15 (54%)||132 (53%)|
| Yes|| 55 (48%)|| 47 (45%)||13 (46%)||115 (47%)|
|Prior or simultaneous radical cystoprostatectomy (n = 244)|
| No|| 95 (85%)|| 91 (88%)||22 (79%)||208 (85%)|
| Yes|| 17 (15%)|| 13 (13%)|| 6 (21%)|| 36 (15%)|
|Suspicious nodes on preoperative CT (n = 209)|
| No|| 82 (83%)|| 69 (81%)||10 (40%)||161 (77%)|
| Yes|| 17 (17%)|| 16 (19%)||15 (60%)|| 48 (23%)|
|BMI median (IQR) (n = 219)|| 26 (23, 29)|| 27 (24, 30)||27 (23, 30)|| 27 (24, 29)|
|Age-adjusted Charlson index median (IQR) (n = 252)|
| || 4 (3, 6)|| 4 (3, 6)|| 4 (2, 5)|| 4 (3, 5)|
|ASA Score (n = 209)|
| 1|| 4 (4%)|| 1 (1%)|| 1 (5%)|| 6 (3%)|
| 2|| 60 (59%)|| 51 (59%)||13 (62%)||124 (59%)|
| 3|| 38 (37%)|| 34 (40%)|| 7 (33%)|| 79 (38%)|
|Number LN dissected median (IQR) (n = 125)|
| ||–|| 4 (2, 10)|| 9 (2, 13)|| 4 (2, 10)|
- Top of page
- Materials and methods
Although the overall benefit of LND for patients undergoing NU for UT-UCC remains unresolved, the potential benefits include accurate pathological assessment of the regional LN. Knowledge of the regional LN status is important because it identifies a high-risk group for recurrence and poor CSS (Fig 2).4,14 Accurate knowledge of the regional LN status (ideally preoperatively) may aid in the appropriate patient selection for adjuvant (or possibly neoadjuvant) therapy including chemotherapy, which is particularly significant given the ongoing progress in the development of non-nephrotoxic chemotherapy agents active against urothelial cancer.15
Despite the recognition that the pathways for progression of UT-UCC include involvement of regional retroperitoneal and pelvic LN, no standards have been defined for when to perform LND or how extensive dissection should be. Hesitancy and lack of standardization still exists even among academic institutions (including our series), where the proportion of patients undergoing any form of LND varies from 44%4 to 67%5 for open NU series.1,2,4,6–12 LND rates are even lower during minimally invasive approaches such as hand-assisted laparoscopic NU9,12 or fully laparoscopic NU,6,12 being approximately 15%. At our institution, LND has become more frequently performed (increasing from 40% to 61% over the last 5 years), yet criteria for its performance still are not uniform. Similarly, it is difficult to draw conclusions regarding local recurrence rates as many series do not report frequency or extent of the LND.6,7,10
Preoperative staging still represents a challenge, and it is not sufficient to identify patients with LNM. Suspicious CT findings was the only variable significantly predicting LNM, similar to reports of Garcia Garcia et al.16 Importantly, 40% of our N+ patients had no sign of LN enlargement on CT.
Tumor size was also not predictive of biological behavior. In our series, two patients with LNM had tumors measuring less than 2 cm. One of them subsequently developed liver metastasis and died of disease. Similarly, patients with low-stage or low-grade tumors are not exempt from developing distant metastasis and death.7,10 None of our patients with LNM had low-grade disease, but one had a Ta tumor. In addition, the pathological variables predicting LNM in multivariate analysis, such as presence of lymphovascular invasion, capsular invasion and sinus fat invasion, are almost impossible to diagnose either on imaging studies or on preoperative biopsy analyses.
Can we select patients that should undergo LND? Given the listed limitations in preoperative staging, we would suggest that proper staging requires LND be performed in all patients with high-grade UT-UCC.
Defining the surgical limits of the LND is important, because the primary lymphatic drainage area of the kidneys and ureters may be extensive and complex, as described by Parker in 1935.17 The lymphatic drainage from the kidneys to regional nodes is somewhat predictable; however, beyond the regional nodes it becomes varied. Of note, malignant disease may disrupt drainage patterns, making the locations of LNM far less predictable. Clearly, either a hilar or a limited LN dissection will not remove all of the primary lymphatic drainage area of the kidneys or ureters.
In addition to its prognostic value, LND may play a therapeutic role for patients with UT-UCC, as suggested by other authors.14,18 Theoretically, a regional LND could remove microscopic regional disease that has not progressed distantly. However, there have been no well designed prospective studies to support the therapeutic value of LND in patients with UT-UCC. Miyake et al.18 reported no significant difference in CSS between patients with and without LND; however, among patients with no lymphovascular invasion identified in the specimen, 5-year CSS was 86% for the 19 patients with LND, significantly higher than the 50% CSS for the 25 patients without LND (P < 0.05). They concluded that lymphadenectomy might provide a therapeutic advantage in patients with UT-UCC in the absence of lymphovascular invasion. In addition, it is not unreasonable to extrapolate results from cystectomy studies, which demonstrated that selected patients with positive LN during cystectomy can obtain survival benefit from extended LND.3 Of note, many patients will have a solitary kidney and thus are not candidates for adjuvant platinum-based chemotherapy; therefore, removal of all identifiable disease may afford the best chance for cure.
The lack of any standardized pattern of side-specific dissection in our cohort represents a limitation in drawing definite conclusions regarding the therapeutic role of LND. Mapping information is also not available because LN were removed in only half of the cohort, with a tendency to perform LN sampling rather than formal LND. While the criteria for performing a LND were difficult to assess retrospectively, it is of interest that survival curves of both Nx and N0 patients were similar. This likely represents an undefined selection bias in that patients with lower risk for LNM were most likely included in a large percentage of the Nx group.
This communication does not pretend to demonstrate the therapeutic benefit of LND, but what it clearly states is the need for therapeutically-focused prospective, randomized protocols that adequately define the techniques and extent of LND in order to find a solution this unsolved urological dilemma.