Urologic Oncology: Seminars and Original Investigations ] (2014) ∎∎∎–∎∎∎
Original article
Contemporary trends in postchemotherapy retroperitoneal lymph node dissection: Additional procedures and perioperative complications Clint Cary, M.D., M.P.H.*, Timothy A. Masterson, M.D., Richard Bihrle, M.D., Richard S. Foster, M.D. Department of Urology, Indiana University School of Medicine, Indianapolis, IN Received 26 June 2014; received in revised form 24 July 2014; accepted 25 July 2014
Abstract Introduction: Postchemotherapy retroperitoneal lymph node dissection (PC-RPLND) is a mainstay in the treatment of men with metastatic testicular cancer. We sought to determine whether trends in the need for additional intraoperative procedures and development of perioperative complications have changed over time. Methods: Patients undergoing PC-RPLND from 2003 to 2011 were identified in the Indiana University Testis Cancer Database. Trends in the incidence of perioperative complications and additional procedures were assessed over time using regression tests of trend. Complications were classified according to the modified Clavien system. Univariable and multivariable logistic regression was used to determine factors associated with undergoing additional procedures. Results: After exclusion criteria, 755 patients were included in the final study cohort. The incidence of additional procedures at PC-RPLND was 22.1% (167 of 755). The rate of additional procedures per year ranged from 17% to 30%, with no significant trend in any direction (Ptrend ¼ 0.66). After adjusting for covariates, preoperative retroperitoneal (RP) mass size, elevated markers, and RP pathology remained significantly associated with the odds of an additional procedure. RP mass size of 410 cm was the strongest predictor (odds ratio ¼ 7.2, 95% CI: 2.6–19.5). Overall, the incidence of perioperative complications was 3.7% (28 of 755). The rate of perioperative complications per year ranged from 0% to 7.3% with no significant trend in any direction (Ptrend ¼ 0.06). Conclusion: The incidence of perioperative complications is low with no significant trend over the last decade. A substantial number of patients require additional intraoperative procedures during PC-RPLND, which has remained stable at our institution over time. r 2014 Elsevier Inc. All rights reserved.
Keywords: Testis cancer; Germ cell tumor; Retroperitoneal lymph node dissection; Complication
1. Introduction In 2014, an estimated 8,820 new cases of testis cancer will be diagnosed in the United States [1]. Fortunately, most of these men will be cured of disease, including those presenting with more advanced disease who require chemotherapy. Those patients with residual retroperitoneal (RP) disease following induction chemotherapy will require Prior presentation: This abstract was presented at the Society of Urologic Oncology 2013 meeting, Bethesda, MD. All funding for this study was provided by the Indiana University Department of Urology. * Corresponding author. Tel.: þ1-317-9-489-272; fax: þ1-317-9-440174. E-mail address: [email protected] (C. Cary). http://dx.doi.org/10.1016/j.urolonc.2014.07.013 1078-1439/r 2014 Elsevier Inc. All rights reserved.
postchemotherapy retroperitoneal lymph node dissection (PC-RPLND). The rationale for retroperitoneal lymph node dissection (RPLND) following chemotherapy is to remove the persistent RP lymph nodes that contain either teratoma or active cancer in 60% of cases [2]. PC-RPLND can be a technically challenging procedure. This is in part owing to the desmoplastic reaction created by systemic chemotherapy in the treatment of metastatic disease, but is also related to the size of the residual RP mass. The desmoplastic tumor response distorts normal tissue planes between the RP tumor and adjacent structures (e.g., aorta, vena cava, kidney, and bowel) making the dissection more demanding, necessitating additional intraoperative procedures and potentially leading to a greater number of perioperative complications.
2
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
Approximately 33% will require an additional procedure (i.e., nephrectomy and bowel resection), whereas up to 5% to 25% of men undergoing RPLND will experience a perioperative complication [3–7]. Over the past 30 years, the radiographic and serologic staging of disease, chemotherapy regimens, surgical techniques, and supportive care for advanced germ cell tumors have been refined. For example, Mosharafa et al. [4] previously demonstrated a trend toward decreasing perioperative complications in the PC-RPLND setting over time comparing 79 men who underwent surgery between 1990 and 1992 with 150 men between 2000 and 2002 (13.9% vs. 6.7%, respectively, P ¼ 0.07). In this study, we sought to evaluate temporal trends in the need for additional intraoperative required at PCRPLND and the development of perioperative complications during the previous decade. As care pathways and staging of disease have improved, we hypothesized a continued decreasing trend in complications and additional procedures would be observed. 1.1. Methods After institutional review board approval, we performed a retrospective cohort study of our prospectively maintained institutional testis cancer database from 2003 to 2011. We identified 870 PC-RPLND cases. All patients underwent an open PC-RPLND. Patients with pure seminoma in the orchiectomy or RP specimen were excluded (n ¼ 38), as these patients are known to be at higher risk of operative complications and incomplete resection [8,9]. Patients undergoing a redo-RPLND (n = 45), and patients lacking baseline clinical and pathologic characteristics (n = 32) were also excluded. 1.1.1. Baseline characteristics Baseline characteristics evaluated included age at surgery, preoperative stage, preoperative mass size, year of surgery, preoperative serum tumor marker status, surgical template, nerve-sparing (yes/no), length of hospital stay, and RP pathology. Patients underwent either a full bilateral template or a modified template as previously described at the discretion of the attending surgeon [10]. Additional procedures performed at the time of PC-RPLND and the development of perioperative complications were determined and prospectively recorded in our institutional database. Additional procedures were defined as any additional surgical procedure that occurred simultaneously during the PC-RPLND (i.e., nephrectomy, hepatic resection, and inferior vena cava resection). Perioperative complication was defined as any complication occurring during admission for the PC-RPLND, which may have prolonged hospital stay. Complications were classified according to the modified Clavien classification system in retrospect by first author (K.C.C.) [11]. To minimize observation bias due to surgeon-related complication misjudgment, a second
coauthor (T.A.M.) independently classified the complications. Any disagreement in coding was resolved by discussion and consensus agreement. This did not include complications that occurred following the date of discharge of the planned procedure due to regional referral patterns with a proportion of patients not returning to our center for follow-up care. Ileus was defined as either requirement of a nasogastric tube postoperatively or delay in the normal clinical course owing to abdominal distention and inability to tolerate diet. 1.1.2. Statistical analysis Temporal trends in the incidence of complications and additional procedures were evaluated using a regression test for trend. Owing to the low frequencies of each type of complication and additional procedure in the individual years, binary variables were created for complication and additional procedure (yes/no). Differences in length of stay between those with and without a complication or additional procedure were tested using the Kruskal-Wallis test. Univariable logistic regression was used to test associations of baseline characteristics with the likelihood of undergoing an additional procedure. Significant variables at the P o 0.10 level in the univariable analysis were used to create the multivariable logistic regression model. Model fit was assessed using the Hosmer-Lemeshow goodness-of-fit test [12]. All analyses were performed using Stata IC version 12 (Stata, College Stattion, TX), with a P r 0.05 considered significant. 2. Results After exclusion criteria, 755 patients were included in the final analysis. Clinical and sociodemographic characteristics are described in Table 1. The median age at diagnosis was 28 years. The site of primary tumor was the testicle in 95.1%. The preoperative RP mass size was 5 to 10 cm in 30.6% and 410 cm in 14%. The average length of stay was 4.4 days. There were 167 (22.1%) additional procedures performed among 135 (17.9%, 135 of 755) patients. The need for additional intraoperative procedures or the development of a perioperative complication was associated with a longer length of hospital stay (P ¼ 0.0001) (Fig. 1). Perioperative complications were identified among 28 (3.7%) patients. Table 2 details the perioperative complications and modified Clavien classification. Among patients not undergoing additional procedures intraoperatively, perioperative complications occurred in 14 patients (2.5%). In comparison, perioperative complications were over 3 times more likely to occur among patients undergoing additional procedures (14 of 176, 8.0%). Of the 28 complications, the most common was postoperative ileus (14%) and deep vein thrombosis (14%). Most complications were classified as grade II (13 of 28, 46%) and grade III (8 of 28, 29%). There were 2 grade IV complications and 2 fatal complications
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7 Table 1 Baseline characteristics of 755 PC-RPLND patients from 2003 to 2011 Characteristic Age, median (IQR)
N 28 (23–36)
Clinical stage at initiation presentation I II III
130 (17.5%) 331 (44.4%) 284 (38.1%)
Site of primary tumor Left testis Right testis Retroperitoneum Other
348 368 25 14
(46.2%) (48.9%) (3.3%) (1.6%)
Preoperative RP mass size r2 cm 2–5 cm 5–10 cm 410 cm
75 344 231 105
(9.9%) (45.6%) (30.6%) (13.9%)
RP histology Necrosis Teratoma Cancer
402 (53.3%) 231 (30.6%) 122 (16.2%)
Length of stay, mean (SD)
4.4 (3.2)
Elevated markers Yes No
181 (23.9%) 574 (76.1%)
Perioperative complications Additional procedures Nephrectomy IVC resection/reconstruction Arterial graft Hepatic resection Duodenal resection Ureterectomy with reconstruction Othera
28 (3.7%) 167 55 28 18 24 7 11 24
(22.1%) (7.3%) (3.7%) (2.4%) (3.2%) (1.0%) (1.5%) (3.2%)
a Cholecystectomy, adrenalectomy, liver biopsy, ileal or colon resection, and mesenteric lymph node biopsy.
(grade V). One patient developed a neurological complication of bilateral lower extremity weakness following a combined RP and thoracic dissection. Overall perioperative mortality was 0.26% (2 of 755). One patient died on postoperative day 2 as a result of a saddle pulmonary embolism. The second perioperative death occurred on postoperative day 13 in a massive teratoma (410 cm) case who required concomitant duodenal resection. Chylous ascites developed with an associated peritoneal infection, abdominal compartment syndrome, and septic shock. Temporal trends of additional procedures remained relatively stable at approximately 20% per year and ranged from 17% to 30% (Fig. 2A). No significant trend of additional procedures was observed during this time period (Ptrend ¼ 0.66). The temporal trends of type of additional procedure are demonstrated in Fig. 2B. Of the 167 additional procedures performed, nephrectomy (33%) was the most common procedure and this ranged from 15% to 50%
3
during the study period. In total, 24 hepatic resections were performed and occurred between 0% and 33% of cases per year that required an additional procedure. In total, 28 perioperative complications occurred with a nonsignificant increasing trend in perioperative complications in the more recent years (Ptrend ¼ 0.06) (Fig. 3). The rate of perioperative complications per year ranged from 0% to 7.3%. Most patients received induction chemotherapy with bleomycin, etoposide, and cisplatin (n ¼ 665, 88%). There were 58 (8%) patients who received etoposide and cisplatin. In total, 30 patients (4%) received VIP (etoposide, ifosfamide, and cisplatin). There was 1 patient who received PVB (ciplatin, vinblastine, and bleomycin) and 1 patient who received high-dose induction chemotherapy outside the United States before being seen at our institution. Salvage chemotherapy was also received by 99 patients before PCRPLND. Preoperative stage, mass size, tumor marker status, type of dissection template (bilateral vs. modified), nerve-sparing procedure, and RP pathology were associated with undergoing an additional procedure in univariable analysis (Table 3). In multivariable logistic regression, the odds of undergoing an additional procedure was 7-fold higher (odds ratio ¼ 7.2, 95% CI: 2.6–19.5) in patients with a preoperative RP mass size of 410 cm compared with those with o2-cm RP mass size. Patients with elevated markers at the time of PC-RPLND had a 90% increased odds of undergoing an additional procedure (odds ratio ¼ 1.9, 95% CI: 1.2–2.9). RP pathology was also associated with undergoing an additional procedure at the time of PC-RPLND (P ¼ 0.003).
3. Discussion PC-RPLND is a challenging procedure requiring additional procedures in up to 25% of patients in this study. The rate of additional procedures has remained stable over time during the previous decade. Perioperative complications occurred in up to 7% patients per year with a nonsignificant increasing trend over time. The current study suggests patients most likely to require an additional procedure at the time of PC surgery have larger RP masses, elevated markers at surgery, and RP histology of cancer or necrosis. The incidence of additional procedures in this series is comparable to previously published reports from other institutions managing this complex patient population, which range from 19% to 33% [6,7,13]. In a German study of 152 patients undergoing PC-RPLND over a 8-year period, the overall incidence of additional procedures was 19.1% [6]. They noted the most common of these procedures were ipsilateral nephrectomy, inferior vena caval resection, and hepatic resection. Djaladat et al. [7] reported that 33% of cases required an additional procedure of 85 PC-RPLND cases performed during a 6-year study period. Nephrectomy and vascular resections were the most
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
4
14 N=14
Mean LOS (days)
12 N=14
12.1
10 8 6 4
10
p=0.0001* N=162 N=556
5
3.9 2 0 None
Additional Complication only Procedure only
Additional Procedure & Complication
*Kruskal-Wallis p-value LOS = length of stay Fig. 1. Association of length of hospital stay with additional procedures and perioperative complications. *Kruskal-Wallis P-value, LOS ¼ length of stay.
common procedure in this study as well. Although the current study suggests a similar incidence of additional procedures at PC-RPLND as previous reports, it adds to the current literature by providing a temporal view over the previous decade demonstrating no significant trends over time in additional procedures. Using a predictive model, Winter et al. [13] demonstrated the likelihood of a vena cava intervention—resection Table 2 Perioperative complications classified according to the modified Clavien system Grade
Complication
N
I
Wound infection
3
II
Pancreatitis SBO Ileus DVT Clostridium difficile Pulmonary embolus
1 1 4 4 2 1
IIIa
Pneumothorax Chest tube placement for pleural effusion Reintubation Vocal cord paralysis
2 2 2 1
IIIb IVa IVb
Small bowel obstruction Neurologica DIC
1 1 1
V
Pulmonary embolus Chylous ascites
1 1
DIC ¼ disseminated intravascular coagulation, DVT ¼ deep vein thrombosis; SBO ¼ small bowel obstruction. a Bilateral lower extremity (LE) weakness following surgery. Started on rehabilitation exercises, and some improvement was observed in LE strength before discharge. Had concomitant RP and mediastinal dissection with division of lumbar arteries. Mass size was 2 to 5 cm in RP.
or cavotomy—increased with RP mass sizes Z5 cm and worse risk category. Indeed, larger RP mass sizes and worse features such as elevated markers and RP histology were also predictive of additional procedures in our study. Our group have previously demonstrated that patients undergoing either caval resection or aortic replacement have more advanced features such as bulky RP disease, elevated markers, or having received salvage chemotherapy [14,15]. Mosharafa et al. [4] previously demonstrated a lower incidence of perioperative complications in a more recent era of surgery (6.7%) compared with a historical era (13.9%), although this was not significant (P ¼ 0.07). The current study expanded on this by evaluating perioperative complications over time during a modern surgical era and use of the modified Clavien classification system to standardize reporting of complications. Complications were as high as 7% per year in the current study, and no significant trend was noted. Overall, perioperative complications were low and this likely reflects increased surgical experience gained over time and standardized postoperative care pathways at a high-volume institution. Institutions with a large experience in treating patients with testicular cancer have previously demonstrated improved outcomes [16]. Indeed, hospitals with higher volumes of RPLND cases have significantly fewer complications compared with lower volume hospitals in a study using data from the Nationwide Inpatient Sample [5]. To ensure that this finding is not unique to testicular cancer, a systematic review by Hillner et al. [17] demonstrated improved surgical outcomes associated with higher volume hospitals and surgeons across numerous cancer types such as colorectal, ovarian, breast, and testicular cancer. We did not observe the hypothesized continued decrease in complications or additional procedures over time in a modern era. However, the comparison of the 3.7%
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
5
Percentage of Patients
100% 18
17
20
2003
2004
2005
16
15
2008
2009
19 26
22
14
2010
2011
80%
60%
40%
20%
0% 2006
2007
Year of Surgery None
Additional
100%
Percentage of Patients
3 6
80% 9
3 7
1
8
6
8
5
4 60%
3
3 40%
5
2
5
2 0 1
2
1 5
5
0 1
1
2
3
0 1
0
0
4
3 1 20%
0% 2003
2004
2005
2006
2007
2008
2009
2010
2011
Year of Surgery Other
Arterial graft
IVC resection
Hepatic resection
Nephrectomy
Other includes: cholecystectomy, adrenalectomy, liver biopsy, ileal and/or colon resection, mesenteric lymph node biopsy etc. Fig. 2. (A) Distribution of any additional procedure by year of surgery in 755 PC-RPLND patients (Ptrend ¼ 0.66). Numbers in each year of surgery represent frequencies of additional procedures. (B) Distribution of type of additional procedure by year of surgery in 755 PC-RPLND patients. Numbers in each year of surgery represent frequencies of each type of additional procedure. (Other: cholecystectomy, adrenalectomy, liver biopsy, ileal or colon resection, and mesenteric lymph node biopsy, etc.)
incidence of complications in this study with the 6.7% and 13.9% incidence of complications in more historical periods from our institution [4] suggested that complications have continued to decrease. It is likely a certain proportion of patients will require intraoperative additional procedures or develop perioperative complications; thus, a certain baseline level should be expected. Therefore, in this health care era
focused on limiting complications and improving the quality of care, we must acknowledge that certain complex surgical patients may develop postoperative complications that simply reflect the nature of the disease rather than the quality of care. There were several significant complications that occurred in the study cohort. A patient developed disseminated
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
6
Percentage of Patients
100%
2
0
3
6
0
3
5
6
3
2009
2010
2011
90%
80%
70%
60%
50% 2003
2004
2005
2006
2007
2008
Year of Surgery None
Complication
Fig. 3. Distribution of perioperative complication by year of surgery in 755 PC-RPLND patients (Ptrend ¼ 0.06). Numbers in each year of surgery represent frequencies of complications.
intravascular coagulation during the operative procedure requiring the procedure to be aborted until this resolved and was ultimately able to undergo an uncomplicated PCRPLND. Another patient developed significant bilateral lower extremity weakness immediately following PCRPLND requiring long-term physical therapy. This patient Table 3 Univariable and multivariable analysis of factors associated with additional procedures in 755 men undergoing PC-RPLND Univariable OR (95% CI)
Multivariable P value
OR (95% CI)
Age 1.01 (0.99–1.03) 0.20 – Preoperative mass size (cm) o2 Referent o0.001 Referent 2–5 1.6 (0.8–3.4) 1.7 (0.7–4.2) 5–10 2.5 (1.2–5.4) 2.4 (0.97–6.2) 410 5.7 (2.6–12.7) 7.2 (2.6–19.5) Preoperative stage I Referent II 2.7 (0.3–21.0) III 6.1 (0.8–48.8)
o0.001 Referent 0.9 (0.1–8.6) 1.5 (0.2–14.7)
P value – o0.001
0.06
Year of surgery Elevated markers Templatea Nerve-sparing
1.02 (0.95–1.09) 0.62 – 3.0 (2.1–4.3) o0.001 1.9 (1.2–2.9) 1.8 (1.3–2.7) 0.001 1.1 (0.7–1.7) 0.50 (0.27–0.92) 0.03 0.9 (0.5–1.7)
– 0.005 0.62 0.71
RP pathology Necrosis Teratoma Cancer
Referent o0.01 0.52 (0.34–0.80) 1.8 (1.2–2.8)
0.003
OR ¼ Odds ratio. Bolded P-values represent significant results. a Bilateral vs. modified unilateral.
Referent 0.4 (0.3–0.7) 1.1 (0.6–1.9)
had a simultaneous RP and retrocrural dissection with division of most of the lumbar arteries in both the abdomen and the mediastinum to facilitate tumor removal. This rare phenomenon has been demonstrated to occur in individuals with extensive dissection including suprahilar dissection [18]. Further, 2 postoperative deaths occurred in this group. One patient (31 y old) died on postoperative day 2 of an acute saddle pulmonary embolus. He had previously been diagnosed with a deep vein thrombosis while receiving his induction chemotherapy. The second patient (27 y old) had a massive RP teratoma requiring the additional procedure of duodenal resection. Lymphatic and pancreatic ascites developed postoperatively and a subsequent infectious complication resulting in septic shock and death on postoperative day 13. Despite a high volume of cases during this study period, several caveats must be acknowledged. First, the identification of perioperative complications was limited to events that occurred during the initial hospital stay of the procedure. Additional postoperative complications, such as chylous ascites, readmission rates, and delayed small bowel obstruction, potentially occurring following discharge were not captured. The nature of being a large referral center with patients presenting from various regions of the country make collection of this postdischarge data challenging. However, collection of this perioperative data is prospectively collected and maintained in our institutional database and is consistent with previous publications from our institution [3,4]. Second, given the frequency with which this procedure is performed at our institution, the results may not be generalizable across practices. Despite these limitations, we feel this study gives an accurate representation of the temporal trends in additional procedures and perioperative complications. Further, features associated
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
with requiring an additional procedure were identified and is useful for physicians treating this population of patients. 4. Conclusions Larger RP mass size, elevated markers, and RP histology were associated with the need for additional procedures during PC-RPLND. The incidence of additional procedures was 23.3% and has remained stable during the previous decade. Perioperative complications associated with this procedure have remained stable. Physicians caring for this patient population should be aware of the possibility for additional surgical procedures to ensure complete resection of disease in this complex patient population. References [1] Siegel R, Ma J, Zou Z, et al. Cancer statistics, 2014. CA Cancer J Clin 2014;64:9–29. [2] Beck SDW, Foster RS. Long-term outcome of retroperitoneal lymph node dissection in the management of testis cancer. World J Urol 2006;24:267–72. [3] Baniel J, Foster RS, Rowland RG, et al. Complications of postchemotherapy retroperitoneal lymph node dissection. J Urol 1995; 153:976–80. [4] Mosharafa AA, Foster RS, Koch MO, et al. Complications of postchemotherapy retroperitoneal lymph node dissection for testis cancer. J Urol 2004;171:1839–41. [5] Yu HY, Hevelone ND, Patel S, et al. Hospital surgical volume, utilization, costs and outcomes of retroperitoneal lymph node dissection for testis cancer. 2012;2012:8, Article ID 189823. Available at: http://dx.doi.org/10.1155/2012/189823. [6] Heidenreich A, Pfister D, Witthuhn R, et al. Postchemotherapy retroperitoneal lymph node dissection in advanced testicular cancer: radical or modified template resection. Eur Urol 2009;55:217–24.
7
[7] Djaladat H, Nichols C, Daneshmand S. Adjuvant surgery in testicular cancer patients undergoing postchemotherapy retroperitoneal lymph node dissection. Ann Surg Oncol 2012;19:2388–93. [8] Flechon A, Bompas E, Biron P, et al. Management of postchemotherapy residual masses in advanced seminoma. J Urol 2002; 168:1975–9. [9] Mosharafa AA, Foster RS, Leibovich BC, et al. Is post-chemotherapy resection of seminomatous elements associated with higher acute morbidity? J Urol 2003;169:2126–8. [10] Jacobsen N-EB, Foster RS, Donohue JP. Retroperitoneal lymph node dissection in testicular cancer. Surg Oncol Clin N Am 2007;16:199– 220. [11] Dindo D, Demartines N, Clavien P-A. Classification of surgical complications. Ann Surg 2004;240:205–13. [12] Hosmer DW Jr, Lemeshow S, Sturdivant RX. Applied Logistic Regression, 3rd ed. Hoboken, New Jersey: Wiley & Sons, 2013. [13] Winter C, Pfister D, Busch J, et al. Residual tumor size and IGCCCG risk classification predict additional vascular procedures in patients with germ cell tumors and residual tumor resection: a multicenter analysis of the German Testicular Cancer Study Group. Eur Urol 2012;61:403–9. [14] Beck SDW, Foster RS, Bihrle R, et al. Aortic replacement during postchemotherapy retroperitoneal lymph node dissection. J Urol 2001; 165:1517–20. [15] Beck SDW, Lalka SG. Long-term results after inferior vena caval resection during retroperitoneal lymphadenectomy for metastatic germ cell cancer. J Vasc Surg 1998;28:808–14. [16] Collette L, Sylvester RJ, Stenning SP, et al. Impact of the treating institution on survival of patients with “poor-prognosis” metastatic nonseminoma. European Organization for Research and Treatment of Cancer Genito-Urinary Tract Cancer Collaborative Group and the Medical Research Council Testicular Cancer Working Party. J Natl Cancer Inst 1999;91:839–46. [17] Hillner BE, Smith TJ, Desch CE. Hospital and physician volume or specialization and outcomes in cancer treatment: importance in quality of cancer care. J Clin Oncol 2000;18:2327–40. [18] Leibovitch I, Nash PA, Little JS, et al. Spinal cord ischemia after post-chemotherapy retroperitoneal lymph node dissection for nonseminomatous germ cell cancer. J Urol 1996;155:947–51.
Original article
Contemporary trends in postchemotherapy retroperitoneal lymph node dissection: Additional procedures and perioperative complications Clint Cary, M.D., M.P.H.*, Timothy A. Masterson, M.D., Richard Bihrle, M.D., Richard S. Foster, M.D. Department of Urology, Indiana University School of Medicine, Indianapolis, IN Received 26 June 2014; received in revised form 24 July 2014; accepted 25 July 2014
Abstract Introduction: Postchemotherapy retroperitoneal lymph node dissection (PC-RPLND) is a mainstay in the treatment of men with metastatic testicular cancer. We sought to determine whether trends in the need for additional intraoperative procedures and development of perioperative complications have changed over time. Methods: Patients undergoing PC-RPLND from 2003 to 2011 were identified in the Indiana University Testis Cancer Database. Trends in the incidence of perioperative complications and additional procedures were assessed over time using regression tests of trend. Complications were classified according to the modified Clavien system. Univariable and multivariable logistic regression was used to determine factors associated with undergoing additional procedures. Results: After exclusion criteria, 755 patients were included in the final study cohort. The incidence of additional procedures at PC-RPLND was 22.1% (167 of 755). The rate of additional procedures per year ranged from 17% to 30%, with no significant trend in any direction (Ptrend ¼ 0.66). After adjusting for covariates, preoperative retroperitoneal (RP) mass size, elevated markers, and RP pathology remained significantly associated with the odds of an additional procedure. RP mass size of 410 cm was the strongest predictor (odds ratio ¼ 7.2, 95% CI: 2.6–19.5). Overall, the incidence of perioperative complications was 3.7% (28 of 755). The rate of perioperative complications per year ranged from 0% to 7.3% with no significant trend in any direction (Ptrend ¼ 0.06). Conclusion: The incidence of perioperative complications is low with no significant trend over the last decade. A substantial number of patients require additional intraoperative procedures during PC-RPLND, which has remained stable at our institution over time. r 2014 Elsevier Inc. All rights reserved.
Keywords: Testis cancer; Germ cell tumor; Retroperitoneal lymph node dissection; Complication
1. Introduction In 2014, an estimated 8,820 new cases of testis cancer will be diagnosed in the United States [1]. Fortunately, most of these men will be cured of disease, including those presenting with more advanced disease who require chemotherapy. Those patients with residual retroperitoneal (RP) disease following induction chemotherapy will require Prior presentation: This abstract was presented at the Society of Urologic Oncology 2013 meeting, Bethesda, MD. All funding for this study was provided by the Indiana University Department of Urology. * Corresponding author. Tel.: þ1-317-9-489-272; fax: þ1-317-9-440174. E-mail address: [email protected] (C. Cary). http://dx.doi.org/10.1016/j.urolonc.2014.07.013 1078-1439/r 2014 Elsevier Inc. All rights reserved.
postchemotherapy retroperitoneal lymph node dissection (PC-RPLND). The rationale for retroperitoneal lymph node dissection (RPLND) following chemotherapy is to remove the persistent RP lymph nodes that contain either teratoma or active cancer in 60% of cases [2]. PC-RPLND can be a technically challenging procedure. This is in part owing to the desmoplastic reaction created by systemic chemotherapy in the treatment of metastatic disease, but is also related to the size of the residual RP mass. The desmoplastic tumor response distorts normal tissue planes between the RP tumor and adjacent structures (e.g., aorta, vena cava, kidney, and bowel) making the dissection more demanding, necessitating additional intraoperative procedures and potentially leading to a greater number of perioperative complications.
2
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
Approximately 33% will require an additional procedure (i.e., nephrectomy and bowel resection), whereas up to 5% to 25% of men undergoing RPLND will experience a perioperative complication [3–7]. Over the past 30 years, the radiographic and serologic staging of disease, chemotherapy regimens, surgical techniques, and supportive care for advanced germ cell tumors have been refined. For example, Mosharafa et al. [4] previously demonstrated a trend toward decreasing perioperative complications in the PC-RPLND setting over time comparing 79 men who underwent surgery between 1990 and 1992 with 150 men between 2000 and 2002 (13.9% vs. 6.7%, respectively, P ¼ 0.07). In this study, we sought to evaluate temporal trends in the need for additional intraoperative required at PCRPLND and the development of perioperative complications during the previous decade. As care pathways and staging of disease have improved, we hypothesized a continued decreasing trend in complications and additional procedures would be observed. 1.1. Methods After institutional review board approval, we performed a retrospective cohort study of our prospectively maintained institutional testis cancer database from 2003 to 2011. We identified 870 PC-RPLND cases. All patients underwent an open PC-RPLND. Patients with pure seminoma in the orchiectomy or RP specimen were excluded (n ¼ 38), as these patients are known to be at higher risk of operative complications and incomplete resection [8,9]. Patients undergoing a redo-RPLND (n = 45), and patients lacking baseline clinical and pathologic characteristics (n = 32) were also excluded. 1.1.1. Baseline characteristics Baseline characteristics evaluated included age at surgery, preoperative stage, preoperative mass size, year of surgery, preoperative serum tumor marker status, surgical template, nerve-sparing (yes/no), length of hospital stay, and RP pathology. Patients underwent either a full bilateral template or a modified template as previously described at the discretion of the attending surgeon [10]. Additional procedures performed at the time of PC-RPLND and the development of perioperative complications were determined and prospectively recorded in our institutional database. Additional procedures were defined as any additional surgical procedure that occurred simultaneously during the PC-RPLND (i.e., nephrectomy, hepatic resection, and inferior vena cava resection). Perioperative complication was defined as any complication occurring during admission for the PC-RPLND, which may have prolonged hospital stay. Complications were classified according to the modified Clavien classification system in retrospect by first author (K.C.C.) [11]. To minimize observation bias due to surgeon-related complication misjudgment, a second
coauthor (T.A.M.) independently classified the complications. Any disagreement in coding was resolved by discussion and consensus agreement. This did not include complications that occurred following the date of discharge of the planned procedure due to regional referral patterns with a proportion of patients not returning to our center for follow-up care. Ileus was defined as either requirement of a nasogastric tube postoperatively or delay in the normal clinical course owing to abdominal distention and inability to tolerate diet. 1.1.2. Statistical analysis Temporal trends in the incidence of complications and additional procedures were evaluated using a regression test for trend. Owing to the low frequencies of each type of complication and additional procedure in the individual years, binary variables were created for complication and additional procedure (yes/no). Differences in length of stay between those with and without a complication or additional procedure were tested using the Kruskal-Wallis test. Univariable logistic regression was used to test associations of baseline characteristics with the likelihood of undergoing an additional procedure. Significant variables at the P o 0.10 level in the univariable analysis were used to create the multivariable logistic regression model. Model fit was assessed using the Hosmer-Lemeshow goodness-of-fit test [12]. All analyses were performed using Stata IC version 12 (Stata, College Stattion, TX), with a P r 0.05 considered significant. 2. Results After exclusion criteria, 755 patients were included in the final analysis. Clinical and sociodemographic characteristics are described in Table 1. The median age at diagnosis was 28 years. The site of primary tumor was the testicle in 95.1%. The preoperative RP mass size was 5 to 10 cm in 30.6% and 410 cm in 14%. The average length of stay was 4.4 days. There were 167 (22.1%) additional procedures performed among 135 (17.9%, 135 of 755) patients. The need for additional intraoperative procedures or the development of a perioperative complication was associated with a longer length of hospital stay (P ¼ 0.0001) (Fig. 1). Perioperative complications were identified among 28 (3.7%) patients. Table 2 details the perioperative complications and modified Clavien classification. Among patients not undergoing additional procedures intraoperatively, perioperative complications occurred in 14 patients (2.5%). In comparison, perioperative complications were over 3 times more likely to occur among patients undergoing additional procedures (14 of 176, 8.0%). Of the 28 complications, the most common was postoperative ileus (14%) and deep vein thrombosis (14%). Most complications were classified as grade II (13 of 28, 46%) and grade III (8 of 28, 29%). There were 2 grade IV complications and 2 fatal complications
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7 Table 1 Baseline characteristics of 755 PC-RPLND patients from 2003 to 2011 Characteristic Age, median (IQR)
N 28 (23–36)
Clinical stage at initiation presentation I II III
130 (17.5%) 331 (44.4%) 284 (38.1%)
Site of primary tumor Left testis Right testis Retroperitoneum Other
348 368 25 14
(46.2%) (48.9%) (3.3%) (1.6%)
Preoperative RP mass size r2 cm 2–5 cm 5–10 cm 410 cm
75 344 231 105
(9.9%) (45.6%) (30.6%) (13.9%)
RP histology Necrosis Teratoma Cancer
402 (53.3%) 231 (30.6%) 122 (16.2%)
Length of stay, mean (SD)
4.4 (3.2)
Elevated markers Yes No
181 (23.9%) 574 (76.1%)
Perioperative complications Additional procedures Nephrectomy IVC resection/reconstruction Arterial graft Hepatic resection Duodenal resection Ureterectomy with reconstruction Othera
28 (3.7%) 167 55 28 18 24 7 11 24
(22.1%) (7.3%) (3.7%) (2.4%) (3.2%) (1.0%) (1.5%) (3.2%)
a Cholecystectomy, adrenalectomy, liver biopsy, ileal or colon resection, and mesenteric lymph node biopsy.
(grade V). One patient developed a neurological complication of bilateral lower extremity weakness following a combined RP and thoracic dissection. Overall perioperative mortality was 0.26% (2 of 755). One patient died on postoperative day 2 as a result of a saddle pulmonary embolism. The second perioperative death occurred on postoperative day 13 in a massive teratoma (410 cm) case who required concomitant duodenal resection. Chylous ascites developed with an associated peritoneal infection, abdominal compartment syndrome, and septic shock. Temporal trends of additional procedures remained relatively stable at approximately 20% per year and ranged from 17% to 30% (Fig. 2A). No significant trend of additional procedures was observed during this time period (Ptrend ¼ 0.66). The temporal trends of type of additional procedure are demonstrated in Fig. 2B. Of the 167 additional procedures performed, nephrectomy (33%) was the most common procedure and this ranged from 15% to 50%
3
during the study period. In total, 24 hepatic resections were performed and occurred between 0% and 33% of cases per year that required an additional procedure. In total, 28 perioperative complications occurred with a nonsignificant increasing trend in perioperative complications in the more recent years (Ptrend ¼ 0.06) (Fig. 3). The rate of perioperative complications per year ranged from 0% to 7.3%. Most patients received induction chemotherapy with bleomycin, etoposide, and cisplatin (n ¼ 665, 88%). There were 58 (8%) patients who received etoposide and cisplatin. In total, 30 patients (4%) received VIP (etoposide, ifosfamide, and cisplatin). There was 1 patient who received PVB (ciplatin, vinblastine, and bleomycin) and 1 patient who received high-dose induction chemotherapy outside the United States before being seen at our institution. Salvage chemotherapy was also received by 99 patients before PCRPLND. Preoperative stage, mass size, tumor marker status, type of dissection template (bilateral vs. modified), nerve-sparing procedure, and RP pathology were associated with undergoing an additional procedure in univariable analysis (Table 3). In multivariable logistic regression, the odds of undergoing an additional procedure was 7-fold higher (odds ratio ¼ 7.2, 95% CI: 2.6–19.5) in patients with a preoperative RP mass size of 410 cm compared with those with o2-cm RP mass size. Patients with elevated markers at the time of PC-RPLND had a 90% increased odds of undergoing an additional procedure (odds ratio ¼ 1.9, 95% CI: 1.2–2.9). RP pathology was also associated with undergoing an additional procedure at the time of PC-RPLND (P ¼ 0.003).
3. Discussion PC-RPLND is a challenging procedure requiring additional procedures in up to 25% of patients in this study. The rate of additional procedures has remained stable over time during the previous decade. Perioperative complications occurred in up to 7% patients per year with a nonsignificant increasing trend over time. The current study suggests patients most likely to require an additional procedure at the time of PC surgery have larger RP masses, elevated markers at surgery, and RP histology of cancer or necrosis. The incidence of additional procedures in this series is comparable to previously published reports from other institutions managing this complex patient population, which range from 19% to 33% [6,7,13]. In a German study of 152 patients undergoing PC-RPLND over a 8-year period, the overall incidence of additional procedures was 19.1% [6]. They noted the most common of these procedures were ipsilateral nephrectomy, inferior vena caval resection, and hepatic resection. Djaladat et al. [7] reported that 33% of cases required an additional procedure of 85 PC-RPLND cases performed during a 6-year study period. Nephrectomy and vascular resections were the most
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
4
14 N=14
Mean LOS (days)
12 N=14
12.1
10 8 6 4
10
p=0.0001* N=162 N=556
5
3.9 2 0 None
Additional Complication only Procedure only
Additional Procedure & Complication
*Kruskal-Wallis p-value LOS = length of stay Fig. 1. Association of length of hospital stay with additional procedures and perioperative complications. *Kruskal-Wallis P-value, LOS ¼ length of stay.
common procedure in this study as well. Although the current study suggests a similar incidence of additional procedures at PC-RPLND as previous reports, it adds to the current literature by providing a temporal view over the previous decade demonstrating no significant trends over time in additional procedures. Using a predictive model, Winter et al. [13] demonstrated the likelihood of a vena cava intervention—resection Table 2 Perioperative complications classified according to the modified Clavien system Grade
Complication
N
I
Wound infection
3
II
Pancreatitis SBO Ileus DVT Clostridium difficile Pulmonary embolus
1 1 4 4 2 1
IIIa
Pneumothorax Chest tube placement for pleural effusion Reintubation Vocal cord paralysis
2 2 2 1
IIIb IVa IVb
Small bowel obstruction Neurologica DIC
1 1 1
V
Pulmonary embolus Chylous ascites
1 1
DIC ¼ disseminated intravascular coagulation, DVT ¼ deep vein thrombosis; SBO ¼ small bowel obstruction. a Bilateral lower extremity (LE) weakness following surgery. Started on rehabilitation exercises, and some improvement was observed in LE strength before discharge. Had concomitant RP and mediastinal dissection with division of lumbar arteries. Mass size was 2 to 5 cm in RP.
or cavotomy—increased with RP mass sizes Z5 cm and worse risk category. Indeed, larger RP mass sizes and worse features such as elevated markers and RP histology were also predictive of additional procedures in our study. Our group have previously demonstrated that patients undergoing either caval resection or aortic replacement have more advanced features such as bulky RP disease, elevated markers, or having received salvage chemotherapy [14,15]. Mosharafa et al. [4] previously demonstrated a lower incidence of perioperative complications in a more recent era of surgery (6.7%) compared with a historical era (13.9%), although this was not significant (P ¼ 0.07). The current study expanded on this by evaluating perioperative complications over time during a modern surgical era and use of the modified Clavien classification system to standardize reporting of complications. Complications were as high as 7% per year in the current study, and no significant trend was noted. Overall, perioperative complications were low and this likely reflects increased surgical experience gained over time and standardized postoperative care pathways at a high-volume institution. Institutions with a large experience in treating patients with testicular cancer have previously demonstrated improved outcomes [16]. Indeed, hospitals with higher volumes of RPLND cases have significantly fewer complications compared with lower volume hospitals in a study using data from the Nationwide Inpatient Sample [5]. To ensure that this finding is not unique to testicular cancer, a systematic review by Hillner et al. [17] demonstrated improved surgical outcomes associated with higher volume hospitals and surgeons across numerous cancer types such as colorectal, ovarian, breast, and testicular cancer. We did not observe the hypothesized continued decrease in complications or additional procedures over time in a modern era. However, the comparison of the 3.7%
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
5
Percentage of Patients
100% 18
17
20
2003
2004
2005
16
15
2008
2009
19 26
22
14
2010
2011
80%
60%
40%
20%
0% 2006
2007
Year of Surgery None
Additional
100%
Percentage of Patients
3 6
80% 9
3 7
1
8
6
8
5
4 60%
3
3 40%
5
2
5
2 0 1
2
1 5
5
0 1
1
2
3
0 1
0
0
4
3 1 20%
0% 2003
2004
2005
2006
2007
2008
2009
2010
2011
Year of Surgery Other
Arterial graft
IVC resection
Hepatic resection
Nephrectomy
Other includes: cholecystectomy, adrenalectomy, liver biopsy, ileal and/or colon resection, mesenteric lymph node biopsy etc. Fig. 2. (A) Distribution of any additional procedure by year of surgery in 755 PC-RPLND patients (Ptrend ¼ 0.66). Numbers in each year of surgery represent frequencies of additional procedures. (B) Distribution of type of additional procedure by year of surgery in 755 PC-RPLND patients. Numbers in each year of surgery represent frequencies of each type of additional procedure. (Other: cholecystectomy, adrenalectomy, liver biopsy, ileal or colon resection, and mesenteric lymph node biopsy, etc.)
incidence of complications in this study with the 6.7% and 13.9% incidence of complications in more historical periods from our institution [4] suggested that complications have continued to decrease. It is likely a certain proportion of patients will require intraoperative additional procedures or develop perioperative complications; thus, a certain baseline level should be expected. Therefore, in this health care era
focused on limiting complications and improving the quality of care, we must acknowledge that certain complex surgical patients may develop postoperative complications that simply reflect the nature of the disease rather than the quality of care. There were several significant complications that occurred in the study cohort. A patient developed disseminated
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
6
Percentage of Patients
100%
2
0
3
6
0
3
5
6
3
2009
2010
2011
90%
80%
70%
60%
50% 2003
2004
2005
2006
2007
2008
Year of Surgery None
Complication
Fig. 3. Distribution of perioperative complication by year of surgery in 755 PC-RPLND patients (Ptrend ¼ 0.06). Numbers in each year of surgery represent frequencies of complications.
intravascular coagulation during the operative procedure requiring the procedure to be aborted until this resolved and was ultimately able to undergo an uncomplicated PCRPLND. Another patient developed significant bilateral lower extremity weakness immediately following PCRPLND requiring long-term physical therapy. This patient Table 3 Univariable and multivariable analysis of factors associated with additional procedures in 755 men undergoing PC-RPLND Univariable OR (95% CI)
Multivariable P value
OR (95% CI)
Age 1.01 (0.99–1.03) 0.20 – Preoperative mass size (cm) o2 Referent o0.001 Referent 2–5 1.6 (0.8–3.4) 1.7 (0.7–4.2) 5–10 2.5 (1.2–5.4) 2.4 (0.97–6.2) 410 5.7 (2.6–12.7) 7.2 (2.6–19.5) Preoperative stage I Referent II 2.7 (0.3–21.0) III 6.1 (0.8–48.8)
o0.001 Referent 0.9 (0.1–8.6) 1.5 (0.2–14.7)
P value – o0.001
0.06
Year of surgery Elevated markers Templatea Nerve-sparing
1.02 (0.95–1.09) 0.62 – 3.0 (2.1–4.3) o0.001 1.9 (1.2–2.9) 1.8 (1.3–2.7) 0.001 1.1 (0.7–1.7) 0.50 (0.27–0.92) 0.03 0.9 (0.5–1.7)
– 0.005 0.62 0.71
RP pathology Necrosis Teratoma Cancer
Referent o0.01 0.52 (0.34–0.80) 1.8 (1.2–2.8)
0.003
OR ¼ Odds ratio. Bolded P-values represent significant results. a Bilateral vs. modified unilateral.
Referent 0.4 (0.3–0.7) 1.1 (0.6–1.9)
had a simultaneous RP and retrocrural dissection with division of most of the lumbar arteries in both the abdomen and the mediastinum to facilitate tumor removal. This rare phenomenon has been demonstrated to occur in individuals with extensive dissection including suprahilar dissection [18]. Further, 2 postoperative deaths occurred in this group. One patient (31 y old) died on postoperative day 2 of an acute saddle pulmonary embolus. He had previously been diagnosed with a deep vein thrombosis while receiving his induction chemotherapy. The second patient (27 y old) had a massive RP teratoma requiring the additional procedure of duodenal resection. Lymphatic and pancreatic ascites developed postoperatively and a subsequent infectious complication resulting in septic shock and death on postoperative day 13. Despite a high volume of cases during this study period, several caveats must be acknowledged. First, the identification of perioperative complications was limited to events that occurred during the initial hospital stay of the procedure. Additional postoperative complications, such as chylous ascites, readmission rates, and delayed small bowel obstruction, potentially occurring following discharge were not captured. The nature of being a large referral center with patients presenting from various regions of the country make collection of this postdischarge data challenging. However, collection of this perioperative data is prospectively collected and maintained in our institutional database and is consistent with previous publications from our institution [3,4]. Second, given the frequency with which this procedure is performed at our institution, the results may not be generalizable across practices. Despite these limitations, we feel this study gives an accurate representation of the temporal trends in additional procedures and perioperative complications. Further, features associated
C. Cary et al. / Urologic Oncology: Seminars and Original Investigations ] (2014) 1–7
with requiring an additional procedure were identified and is useful for physicians treating this population of patients. 4. Conclusions Larger RP mass size, elevated markers, and RP histology were associated with the need for additional procedures during PC-RPLND. The incidence of additional procedures was 23.3% and has remained stable during the previous decade. Perioperative complications associated with this procedure have remained stable. Physicians caring for this patient population should be aware of the possibility for additional surgical procedures to ensure complete resection of disease in this complex patient population. References [1] Siegel R, Ma J, Zou Z, et al. Cancer statistics, 2014. CA Cancer J Clin 2014;64:9–29. [2] Beck SDW, Foster RS. Long-term outcome of retroperitoneal lymph node dissection in the management of testis cancer. World J Urol 2006;24:267–72. [3] Baniel J, Foster RS, Rowland RG, et al. Complications of postchemotherapy retroperitoneal lymph node dissection. J Urol 1995; 153:976–80. [4] Mosharafa AA, Foster RS, Koch MO, et al. Complications of postchemotherapy retroperitoneal lymph node dissection for testis cancer. J Urol 2004;171:1839–41. [5] Yu HY, Hevelone ND, Patel S, et al. Hospital surgical volume, utilization, costs and outcomes of retroperitoneal lymph node dissection for testis cancer. 2012;2012:8, Article ID 189823. Available at: http://dx.doi.org/10.1155/2012/189823. [6] Heidenreich A, Pfister D, Witthuhn R, et al. Postchemotherapy retroperitoneal lymph node dissection in advanced testicular cancer: radical or modified template resection. Eur Urol 2009;55:217–24.
7
[7] Djaladat H, Nichols C, Daneshmand S. Adjuvant surgery in testicular cancer patients undergoing postchemotherapy retroperitoneal lymph node dissection. Ann Surg Oncol 2012;19:2388–93. [8] Flechon A, Bompas E, Biron P, et al. Management of postchemotherapy residual masses in advanced seminoma. J Urol 2002; 168:1975–9. [9] Mosharafa AA, Foster RS, Leibovich BC, et al. Is post-chemotherapy resection of seminomatous elements associated with higher acute morbidity? J Urol 2003;169:2126–8. [10] Jacobsen N-EB, Foster RS, Donohue JP. Retroperitoneal lymph node dissection in testicular cancer. Surg Oncol Clin N Am 2007;16:199– 220. [11] Dindo D, Demartines N, Clavien P-A. Classification of surgical complications. Ann Surg 2004;240:205–13. [12] Hosmer DW Jr, Lemeshow S, Sturdivant RX. Applied Logistic Regression, 3rd ed. Hoboken, New Jersey: Wiley & Sons, 2013. [13] Winter C, Pfister D, Busch J, et al. Residual tumor size and IGCCCG risk classification predict additional vascular procedures in patients with germ cell tumors and residual tumor resection: a multicenter analysis of the German Testicular Cancer Study Group. Eur Urol 2012;61:403–9. [14] Beck SDW, Foster RS, Bihrle R, et al. Aortic replacement during postchemotherapy retroperitoneal lymph node dissection. J Urol 2001; 165:1517–20. [15] Beck SDW, Lalka SG. Long-term results after inferior vena caval resection during retroperitoneal lymphadenectomy for metastatic germ cell cancer. J Vasc Surg 1998;28:808–14. [16] Collette L, Sylvester RJ, Stenning SP, et al. Impact of the treating institution on survival of patients with “poor-prognosis” metastatic nonseminoma. European Organization for Research and Treatment of Cancer Genito-Urinary Tract Cancer Collaborative Group and the Medical Research Council Testicular Cancer Working Party. J Natl Cancer Inst 1999;91:839–46. [17] Hillner BE, Smith TJ, Desch CE. Hospital and physician volume or specialization and outcomes in cancer treatment: importance in quality of cancer care. J Clin Oncol 2000;18:2327–40. [18] Leibovitch I, Nash PA, Little JS, et al. Spinal cord ischemia after post-chemotherapy retroperitoneal lymph node dissection for nonseminomatous germ cell cancer. J Urol 1996;155:947–51.
