The Mayo Clinic experience with surgical management, complications and outcome for patients with renal cell carcinoma and venous tumour thrombus


Michael L. Blute, MD, Department of Urology, Mayo Clinic, 200 First Street South-west, Rochester, Minnesota 55905, USA.


The Mayo clinic experience with renal carcinoma and venous tumour thrombus is presented in this section. The authors show that the surgical management of these patients continues to develop, and that complications and mortality are decreasing. They also show that cancer-specific survival is better with renal vein involvement only, as compared with vena caval involvement.

Authors from Dallas examined whether Gleason 3+4 tumours behaved differently to 4+3, and found that that the latter pattern is more aggressive. They propose that the Gleason 4 pattern deserves further molecular study.

Several authors from the USA and Austria compared the performance of complexed PSA with that of total PSA and percentage free PSA; complexed PSA provided better specificity than the other two tests and reduced the number of unnecessary biopsies in patients with a total PSA of 2.6–4.0 ng/mL.


To report the surgical management, complications and outcomes over three decades by tumour thrombus level for patients with renal cell carcinoma (RCC) and renal venous extension, as surgery is the most effective treatment.


We assessed 540 patients who underwent surgical resection for RCC with renal venous extension between 1970 and 2000. Early and late surgical complications, including operative mortality, were compared with tumour thrombus level using the chi-square, Fisher's exact and Wilcoxon rank-sum tests. Cancer-specific survival was estimated using the Kaplan-Meier method and compared across tumour thrombus levels using log-rank tests.


There were 349 (64.6%) patients with level 0 thrombus and 191 (35.4%) with inferior vena cava thrombus, including 66 (12.2%) with level I, 77 (14.3%) with level II, 28 (5.2%) with level III, and 20 (3.7%) with level IV thrombus. Patients with a higher thrombus level had more early surgical complications (respectively for level 0 to IV, 8.6%, 15.2%, 14.1%, 17.9% and 30.0%, P < 0.001). However, there was no statistically significant difference in the incidence of late complications by thrombus level (P = 0.445). The incidence of any early surgical complication decreased from 13.4% for patients treated in 1970–1989 to 8.1% for those treated in 1990–2000 (P = 0.064); the respective operative mortality decreased from 3.8% to 2.0% (P = 0.260), and in patients with inferior vena cava thrombus, from 8.1% to 3.8% (P = 0.227). The respective duration of hospitalization decreased from a median of 8 to 7 days (P < 0.001) but the incidence of late complications increased significantly over time (P < 0.001.) Among patients with clear cell RCC, the respective estimated 5-year cancer-specific survival rates (Se, number still at risk) for patients with level 0 to IV thrombus were 49.1 (3.0)% (125), 31.7 (6.4)% (14), 26.3 (6.1)% (11), 39.4 (10.7)% (7) and 37.0 (12.9)% (5), (P = 0.028). There was a statistically significant difference in outcome for patients with level 0 vs those with level >0 thrombus (P = 0.002), but there was no significant difference in outcome by thrombus level among patients with inferior vena cava tumour thrombus (P = 0.868).


The surgical management of RCC with renal venous extension continues to develop. The incidence of early surgical complications and operative death have decreased in recent times with the introduction of improved imaging, surgical monitoring and vascular bypass techniques. There is significantly better cancer-specific survival for patients with renal vein involvement only than those with inferior vena cava involvement.


transoesophageal echocardiography


inferior vena cava.


The rationale for the aggressive management of RCC with renal venous extension has been established [1–10]. Historically, up to 10% of patients with RCC have tumour thrombus involving the renal vein or vena cava, and 1% have tumour thrombus extending into the right atrium. Factors that contribute to the reported variability in outcome after surgery for patients with RCC and renal venous extension include the assessment before surgery, completeness of resection and the biological characteristics of the primary tumour. Survival can be significant in the absence of distant metastases or positive regional lymph nodes, with pathological features indicating organ-confined status [11–14]. The 5-year cancer-specific survival is ≈ 60% for the most favourable tumours [5,6,10].

Improvements in cross-sectional imaging, the introduction of dynamic monitoring with trans-oesophageal echocardiography (TEE), and the adoption of vascular bypass techniques, e.g. cardiopulmonary bypass and hypothermic circulatory arrest or venous bypass using extracorporeal membrane oxygenators, have improved the safety of these operations, which remain the most challenging for genitourinary oncological surgeons [15–20]. We report the impact of these advances on our institutional practice in terms of operative planning, surgical morbidity and therapeutic outcome of patients treated with radical nephrectomy and tumour thrombectomy with renal venous extension.


Our institutional database of 2838 patients treated with radical nephrectomy between 1970 and 2000 contains records for 540 (19%) patients with renal venous extension, including 191 (7%) with inferior vena cava (IVC) extension and 20 (1%) with superdiaphragmatic extension. Patients treated with radical nephrectomy for RCC with renal venous extension between 1970 and 2000 were eligible for the study. Patients with bilateral synchronous disease, Von Hippel-Lindau or tuberous sclerosis syndromes, Wilms’ tumour, who were <18 years old at surgery, or who denied access to medical records for research, were excluded. After these exclusions, there were 540 patients available for analysis.


The microscopic slides from all tumour specimens were reviewed by a urological pathologist (J.C.C.) using a BX40 microscope (Olympus, Optical Corporation, Melville, New York), while unaware of patient outcome. The histological subtype was assessed following the 1997 Union Internationale Contre le Cancer and American Joint Committee on Cancer classification of RCC [21–23]. Tumours were staged using the 2003 TNM classification. Nuclear grade was assigned using standardized criteria, as previously defined [24]. Perinephric fat was defined as peripheral fat or pelvic sinus fat. Histological tumour necrosis was defined as the presence of any microscopic coagulative tumour necrosis. Degenerative changes such as hyalinization, haemorrhage and fibrosis were not considered necrosis. A sarcomatoid component was defined as a spindle cell malignancy that had the histological appearance of a sarcoma.


The level of tumour thrombus was classified as 0 (thrombus limited to the renal vein, detected clinically or during assessment of the pathological specimen), I (thrombus extending ≤ 2 cm above the renal vein), II (thrombus extending >2 cm above the renal vein, but below the hepatic veins), III (thrombus at the level of or above the hepatic veins but below the diaphragm), and IV (thrombus extending above the diaphragm).

Regardless of the tumour thrombus level, the most important aspect of resection is that involving the vena cava. As tumours within the vena cava can enlarge rapidly, we rely only on recent radiographic imaging when planning the surgical approach. For patients with extensive vena cava tumour thrombus, arrangements for possible vascular bypass are completed before surgery.

When a bypass is unlikely we prefer an anterior subcostal or midline abdominal incision. When a bypass is indicated we routinely use a midline abdominal median sternotomy incision. We rarely use a thoracoabdominal approach. However, if the patient requires a bypass and previously has undergone coronary artery bypass grafting, we use a right lateral thoracoabdominal incision to access the aortic arch and right atrium.

An important aspect of the operation is early ligation of the renal artery to help decrease the chance of troublesome venous bleeding. This may also reduce the cephalad extent of vena cava tumour thrombus because the major blood supply to the tumour thrombus arises from the renal artery. For right-sided lesions we prefer to ligate the renal artery in the interaortocaval region. Similarly, on very rare occasions, we have used preoperative arterial embolization to shrink the vena cava tumour and aid resection.


As a rule, level I tumours cause only partial vena cava occlusion and can be resected with no extensive vena cava dissection or bypass. The tumour can be milked easily into the renal vein and a vascular clamp applied. The ostium is then opened and excised, the kidney and entire renal vein removed, and the caval defect oversewn with running sutures of 4–0 polypropylene. This technique prevents the cephalad propagation of tumour and allows uninterrupted blood flow through the contralateral renal vein and vena cava.


More extensive vena cava dissection and possibly lumbar vein ligation are necessary for proximal and distal control of level II tumours. Generally, most level II tumours can be resected with no bypass. The critical operative manoeuvre during resection of these intracaval tumours occurs after the kidney has been mobilized and Rummel tourniquets or clamps placed sequentially on the infrarenal vena cava, contralateral renal vein and suprarenal vena cava. When necessary, careful dissection and ligation of small hepatic veins to the caudate lobe will allow retraction and exposure of the intrahepatic vena cava superiorly to just below the major hepatic veins for clamp placement.

After assuring vascular control scissors are used to make a cavotomy from the renal ostium cephalad over the tumour. The caval tumour is gently freed with a dissector, and the ostium and renal vein circumscribed and removed together with the kidney. Bothersome bleeding can sometimes occur from unrecognized lumbar veins.

After removing the vena cava tumour thrombus the lumen is flushed and inspected for residual tumour fragments or irregularities that may require biopsy. The cavotomy is then closed with running sutures of 4–0 polypropylene. As the last suture is tightened the distal clamp is released to remove retained air or clot (or both). The clamps are removed, beginning with the distal IVC, and proceeding to the contralateral kidney and proximal clamp.


Although resection of level III tumours can require vascular bypass we have safely resected these tumours using classical techniques. Intraoperative TEE is routinely used for level III tumours to assess haemodynamics and aid resection. The abdominal procedure is completed as previously described for level II tumours. However, hepatic mobilization to expose the retrohepatic IVC and use of the Pringle manoeuvre to control hepatic inflow are used commonly.

After arterial ligation, the need for vascular bypass is reassessed. In resecting some level III tumours, occlusion of the vena cava can compromise venous return with a subsequent decrease in cardiac output, hypotension and hypoperfusion of vital organs. Cross-clamping can also cause extensive haemorrhage from venous collaterals. Collectively, this situation can contribute to incomplete tumour removal and a hurried reconstruction of the vena cava. Dramatic shifts in haemodynamic factors can also contribute to intraoperative death, renal failure and vital organ injury.

To avoid this situation we routinely observe the effects of cross-clamping before proceeding with cavotomy. In our experience the resection of tumours completely occluding the vena cava is better tolerated. Paradoxically, patients with a partially occluded vena cava have a greater chance of intraoperative bleeding, hypotension and incomplete resection. Veno-venous bypass therefore makes resection safer among patients who have a level III tumour thrombus with a partially occluded vena cava or who cannot tolerate cross-clamping. Veno-venous bypass can be used to facilitate resection of vena cava tumour thrombus extending up to the level of the diaphragm.

Veno-venous bypass is initiated after extensive dissection of the vena cava and ligation of the lumbar veins. Initially, a 20 F venous cannula is introduced through a purse-string suture in the IVC, well below the distal aspect of the tumour thrombus. Cannulation of the vena cava in an area containing tumour or bland thrombus increases the risk of entrailing thrombus to the right heart, with possible catastrophic results. Next, an 8–14 F venous cannula is introduced into the right atrium or into the right brachial vein. Each cannula is connected via modified heparin-bonded Gott aneurysm shunt tubing to an electromagnetic centrifugal pump that returns the effluent to the right atrium or axilla.

While we frequently use veno-venous techniques for level III tumours requiring bypass, cardiopulmonary bypass with circulatory arrest and profound hypothermia (18 °C) can also be used. Additional exposure of the vena cava can be obtained by dividing the triangular and coronary ligaments and rotating the right lobe of the liver to the left. The vena cava is then controlled sequentially, from the distal vena cava to the contralateral renal vein and then the proximal vena cava.

During resection of a right-sided tumour, occlusion of the left renal vein near the vena cava may not be associated with a significant rise in the left renal venous pressure, especially when the vena cava has been completely occluded and the second lumbar vein enters the left renal vein posteriorly (about half of the cases).

After adequate vascular control is obtained, cavotomy and vena caval reconstruction are completed as previously described for level II tumours. In situations where the lumen of the vena cava is compromised but complete resection is unnecessary, we reconstruct the vena cava using synthetic grafts. A pericardial patch can also be considered if a cardiopulmonary bypass was used. While veno-venous and cardiopulmonary bypass facilitate the resection of extensive level III tumours and prevent haemodynamic instability in patients with compromised cardiac function, these procedures can increase operating time and blood loss, and prolong hospitalization. Thus, when the vena cava tumour thrombus can be resected safely with no bypass we prefer to use classical techniques.


We typically use a cardiopulmonary bypass and circulatory arrest during resection of level IV tumour thrombus. However, recent experience with level IV tumour thrombus and venous bypass techniques are considered when TEE reveals free-floating thrombus that can be easily reduced below the diaphragm. Nevertheless, surgical resection remains challenging and requires a coordinated thoracic and abdominal approach. The abdominal portion of the procedure is completed as previously described for level III tumours.


Early complications (<30 days after radical nephrectomy) recorded included death during surgery, or after surgery but during hospitalization, haemorrhage, deep vein thrombosis, pulmonary embolism, myocardial infarction, wound infection, abscess, sepsis, acute renal failure, dialysis, the need for an additional surgery, ileus, pneumothorax, duration of hospitalization, blood loss during surgery, units of blood given during surgery and units of blood given during hospitalization. Late complications (30 days to 1 year after radical nephrectomy) included chronic renal insufficiency (creatinine >20 mg/L), proteinuria (protein osmolality ratio >0.12), wound hernia, chronic renal failure, and dialysis.


Surgical features and complications by thrombus level were compared using the chi-square, Fisher's exact, Wilcoxon rank-sum and Kruskal–Wallis tests. Cancer-specific survival for patients with clear cell RCC was estimated using the Kaplan-Meier method. The duration of follow-up was calculated from the date of radical nephrectomy to the date of death or last follow-up. Deaths from causes other than RCC were censored. Comparisons of outcome were evaluated with log-rank tests; in all tests P < 0.05 was considered to indicate statistical significance.


There were 349 (64.6%) patients with level 0 thrombus and 191 (35.4%) with IVC tumour thrombus, including 66 (12.2%) with level I, 77 (14.3%) with level II, 28 (5.2%) with level III, and 20 (3.7%) with level IV tumour thrombus.

The surgical characteristics of patients with IVC tumour thrombus are summarized in Table 1. Operative and anaesthesia times were significantly lower for patients with veno-venous bypass than for those with cardiopulmonary bypass; the respective median (range) operative duration was 265 (217–318) min vs 376 (300–535) min (P = 0.002), and the respective anaesthesia times were 346 (285–385) and 441  (355–568) min (P = 0.004).

Table 1.  Surgical characteristics by tumour thrombus level for patients with IVC thrombus
CharacteristicTumour thrombus level
  1. CPB, cardiopulmonary bypass; CA, circulatory arrest.

N (%)6677 28 20
Intraoperative TEE 0 1 (1.3)  3 (10.7)  11 (55)
Type of incision
Abdominal66 (100)68 (88.3) 20 (71.4)  0
Thoracoabdominal 0 9 (11.7)  8 (28.6)  0
Abdominal/mediastinal 0 0  0 20 (100)
Resection of vena cavaNA 3 (3.9)  1 (3.6) 10 (50)
Occluded vena cava
Partial66 (100)59 (76.6) 18 (64.3) 20 (100)
Complete 018 (23.4) 10 (35.7)  0
Vena cava clips used 0 6 (7.8)  5 (17.9)  6 (30)
Liver mobilized 0 6 (7.8)  7 (25.0)  0
Pringle manoeuvre 0 6 (7.8)  5 (17.9)  0
Extracorporeal circulation
None66 (100)74 (96.1) 20 (71.4)  0
CPB 0 0  0  1 (5)
CPB/CA 0 1 (1.3)  6 (21.4) 13 (65)
Venous 0 2 (2.6)  2 (7.1)  6 (30)
Median (range) duration (min) of
bypassNANA 47.5 (20–123) 87 (11–162)
CANA23 (15–35) 18 (15–28) 18 (13–38)
operationNANA276 (186–497)345 (217–535)
anaesthesiaNANA373 (230–548)396 (285–568)

Four patients (14%) among the 28 with level II–IV tumour thrombus and complete occlusion of the vena cava required a bypass, compared with 27 (28%) of the 97 with partial occlusion (P = 0.144); respectively, six (21%) had at least one early complication, compared with 16 (16%) (P = 0.577), and eight (29%) had at least one late complication, compared with 28 (29%) (P = 0.976).

Six (29%) of the 21 patients with level II–IV tumour thrombus and with a cardiopulmonary bypass had at least one early complication, compared with one of the 10 patients with veno-venous bypass (P = 0.379); respectively six (29%) had at least one late complication compared with two (P = 0.610).


The early and late surgical complications are summarized by tumour thrombus level in Table 2. Seven patients died during surgery (1.3%) and 10 peri-operatively (1.9%), for a total operative mortality of 3.2%. The former were caused by cardiac arrest in three patients, tumour embolization in two and haemorrhagic coagulopathy in two. There was a statistically significant difference in the incidence of any early surgical complication by tumour thrombus level (P = 0.016), with patients with a higher thrombus level having more complications (Table 2). There was no statistically significant difference in the duration of hospitalization by tumour thrombus level (P = 0.403), although there was evidence that patients with level IV thrombus had longer hospital stays than those with level < IV (median 9 vs 8 days, P = 0.059). There were significant increases in blood loss during surgery, units of blood given during surgery and units of blood given during hospitalization as the tumour thrombus level increased (P < 0.001), but there was no statistically significant difference in the incidence of late complications by thrombus level (P = 0.445, Table 2).

Table 2.  Complications and histological subtype by tumour thrombus level
ComplicationTumour thrombus level
  • *

    P = 0.016,

  • P < 0.001.

Early (< 30 days)
Intraoperative death  2 (0.6) 0 3 (3.9) 2 (7.1) 0
Perioperative death  4 (1.2) 1 (1.5) 0 2 (7.1) 3 (15)
Haemorrhage  3 (0.9) 1 (1.5) 2 (2.6) 3 (10.7) 5 (25)
Deep vein thrombosis  2 (0.6) 2 (3.0) 1 (1.3) 0 0
Pulmonary embolism  5 (1.4) 3 (4.6) 2 (2.6) 1 (3.6) 0
Myocardial infarction  2 (0.6) 1 (1.5) 1 (1.3) 2 (7.1) 0
Wound infection  8 (2.3) 1 (1.5) 2 (2.6) 0 0
Abscess  0 1 (1.5) 0 0 1 (5)
Sepsis  1 (0.3) 0 0 0 0
Acute renal failure  1 (0.3) 1 (1.5) 1 (1.3) 1 (3.6) 0
Dialysis  0 1 (1.5) 0 1 (3.6) 0
Additional surgery  9 (2.6) 2 (3.0) 3 (3.9) 1 (3.6) 3 (15)
Ileus  2 (0.6) 1 (1.5) 0 1 (3.6) 0
Pneumothorax  0 2 (3.0) 0 0 0
Any* 30 (8.6)10 (15.2) 11 (14.3) 5 (17.9) 6 (30)
Median (range):
Hospitalization, days  8 (1–36) 7 (4–28) 7 (1–30) 8 (1–28) 9 (1–31)
Blood loss during surgery, L  0.6 (0.05–12.0) 1.0 (0.15–6.0) 1.3 (0.2–8.0) 2.7 (0.6–15.0) 2.5 (0.5–4.0)
Units of blood:
during surgery  1 (0–21) 2 (0–16) 3 (0–35) 6.5 (0–36) 6.5 (0–36)
during hospitalization  2 (0–23) 2 (0–18) 4 (0–35) 9 (0–46) 11.5 (0–46)
Late (30 days to 1 year)
Chronic renal insufficiency 22 (6.3) 9 (13.6) 11 (14.3) 3 (10.7) 1 (5)
Proteinuria 55 (15.8) 8 (12.1)12 (15.6) 4 (14.3) 4 (20)
Wound hernia  0 0 0 0 0
Chronic renal failure  6 (1.7) 0 0 0 0
Dialysis  4 (1.2) 0 1 (1.3) 0 0
Other unspecified 12 (3.4) 3 (4.6) 5 (6.5) 0 0
Any 78 (22.4)17 (25.8)25 (32.5) 6 (21.4) 5 (25)
RCC histological subtype
Clear cell331 (94.8)60 (90.9)68 (88.3)25 (89.3)19 (95)
Papillary  8 (2.3) 2 (3.0) 6 (7.8) 2 (7.1) 1 (5)
Chromophobe  7 (2.0) 0 1 (1.3) 1 (3.6) 0
Collecting duct  0 2 (3.0) 1 (1.3) 0 0
Purely sarcomatoid  1 (0.3) 0 0 0 0
Not otherwise specified  2 (0.6) 2 (3.0) 1 (1.3) 0 0


There was a significant change in the distribution of tumour thrombus level with time (P < 0.001, Table 3). The early and late surgical complications are also summarized with time in Table 3. The incidence of any early surgical complication decreased between the periods assessed, but not significantly (P = 0.064), as did operative mortality, from 3.8% in 1970–1989 to 2.0% in 1990–2000 (P = 0.260). Among patients with IVC tumour thrombus, operative mortality decreased from 8.1% to 3.8% between the periods (P = 0.227). The duration of hospitalization decreased significantly but the incidence of late complications increased significantly (both P < 0.001, Table 3).

Table 3.  Thrombus level and surgical complications (early and late) by decade of nephrectomy
Decade of nephrectomy1970–891990–2000
  • *

    P < 0.001

Thrombus level
0257 (74.9) 92 (46.7)
I 31 (9.0) 35 (17.8)
II 39 (11.4) 38 (19.3)
III 12 (3.5) 16 (8.1)
IV  4 (1.2) 16 (8.1)
Early complications
Death during surgery  6 (1.8)  1 (0.5)
Perioperative death  7 (2.0)  3 (1.5)
Haemorrhage  9 (2.6)  5 (2.5)
Deep vein thrombosis  3 (0.9)  2 (1.0)
Pulmonary embolism  11 (3.2)  0
Myocardial infarction  5 (1.5)  1 (0.5)
Wound Infection 10 (2.9)  1 (0.5)
Abscess  1 (0.3)  1 (0.5)
Sepsis  1 (0.3)  0
Acute renal failure  3 (0.9)  1 (0.5)
Dialysis  1 (0.3)  1 (0.5)
Additional surgery 12 (3.5)  6 (3.1)
Ileus  2 (0.6)  2 (1.0)
Pneumothorax  1 (0.3)  1 (0.5)
Any 46 (13.4) 16 (8.1)
Median (range):
hospitalization, days*  8 (1−35)  7 (1−36)
Blood loss during surgery, L  0.8 (0.05−15.0)  0.75 (0.05−7.4)
Units of blood
during surgery  2 (0−36)  2 (0−35)
during hospitalization  2 (0−46)  2 (0−40)
Late complications
Chronic renal insufficiency 26 (7.6) 20 (10.2)
Proteinuria 28 (8.2) 55 (27.9)
Wound hernia  0  0
Chronic renal failure  2 (0.6)  4 (2.0)
Dialysis  1 (0.3)  4 (2.0)
Other unspecified 19 (5.5)  1 (0.5)
Any* 62 (18.1) 69 (35.0)


Among the 540 patients studied, there were 503 (93.2%) with clear cell RCC, 19 (3.5%) with papillary RCC, nine (1.7%) with chromophobe RCC, three (0.6%) with collecting duct RCC, one (0.2%) with purely sarcomatoid RCC and no other underlying histological subtype, and five (0.9%) with RCC ‘not otherwise specified’. There was no statistically significant association between RCC histological subtype and thrombus level (P = 0.132, Table 2).


The 2003 TNM stage, nuclear grade, histological tumour necrosis and presence of a sarcomatoid component for the 503 patients with clear cell RCC are summarized in Table 4. Of the 503 patients with clear cell RCC, 417 died, including 311 from RCC. The mean (median, range) time from radical nephrectomy to death from RCC was 3.1 (1.7, 0–25) years. The mean time from radical nephrectomy to the last follow-up for the 86 patients still alive at the last follow-up was 10.3 (8.4, 0–30) years. Cancer-specific survival by tumour thrombus level for patients with clear cell RCC is shown in Fig. 1A. There was a statistically significant difference in outcome by thrombus level (P = 0.028), although this reflected the difference in outcome for patients with level 0 thrombus compared with patients with level >0 thrombus (P = 0.002, Fig. 1B). There was no statistically significant difference in outcome by thrombus level among patients with IVC tumour thrombus (P = 0.868). Table 4 also summarizes the 5-year cancer-specific survival rates and comparisons of outcome among the pathological features of interest. Among the combinations of regional lymph node involvement and distant metastases (Fig. 1C), cancer-specific survival was greatest (59%) for patients with pN0/pNx, pM0 disease and poorest for patients with pN1/pN2, pM1 disease (5.8%).

Table 4.  Pathological features and cancer-specific survival for 503 patients with clear cell RCC
FeatureN (%)5-year CSS (Se, n at risk)P
  1. CSS, cancer-specific survival.

2003 Primary tumour stage
pT3b470 (93.4)44.4 (2.5, 156)<0.001
pT3c 17 (3.4)43.3 (14.3, 5) 
pT4 16 (3.2) 6.8 (6.5, 1) 
Perinephric fat invasion
No192 (38.2)56.2 (3.8, 86)<0.001
Yes272 (54.1)32.3 (3.1, 61) 
Regional lymph node involvement
pNx/pN0449 (89.3)47.1 (2.6, 156)<0.001
pN1/pN2 54 (10.7)12.4 (4.7, 6) 
Distant metastases
pM0361 (71.8)55.3 (2.9, 146)<0.001
pM1142 (28.2)13.2 (3.0, 16) 
Combination of N and M stage
pN0/pNx, pM0332 (66.0)59.1 (3.0, 141)<0.001
pN1/pN2, pM0 29 (5.8)17.2 (7.0, 5) 
pN0/pNx, pM1 117 (23.3)14.7 (3.5, 15) 
pN1/pN2, pM1 25 (5.0) 5.8 (5.5, 1) 
Nuclear grade
1 14 (2.8)92.3 (7.4, 12)<0.001
2105 (20.9)65.5 (5.0, 55) 
3304 (60.4)40.3 (3.1, 88) 
4 80 (15.9)12.2 (4.2, 7) 
Histological tumour necrosis
No243 (48.3)61.3 (3.4, 114)<0.001
Yes260 (51.7)25.6 (3.0, 48) 
Sarcomatoid component
No457 (90.9)46.7 (2.5, 161)<0.001
Yes 46 (9.2) 3.3 (3.2, 1) 
Figure 1.

The estimated cancer-specific survival rates at 5 years for patients with: A, level 0, I, II, III, and IV tumour thrombus; (Se) (number still at risk) rates were 49.1 (3.0)% (125), 31.7 (6.4)% (14), 26.3 (6.1)% (11), 39.4 (10.7)% (7) and 37.0 (12.9)% (5), respectively (P = 0.028); B, with level 0 and > level 0 tumour thrombus (rate 31.5%, Se 3.9%, 37) (P = 0.002); C, pNX/pN0 pM0, pN1/pN2 pM0, pNX/pN0 pM1, and pN1/pN2 pM1 RCC; the rates 59.1 (3.0)% (141), 17.2 (7.0)% (5), 14.7 (3.5)% (15) and 5.8 (5.5)% (1), respectively (P < 0.001).


RCC extending to the renal vein, vena cava or right atrium presents a challenging surgical management problem. Previous reports show that aggressive surgical resection of these extensive lesions can produce long-term freedom from disease [1–10]. In addition, multivariate analyses showed that primary tumour characteristics, e.g. tumour stage, grade, perinephric fat invasion, lymph node involvement, or presence of distant metastases, determine the outcome rather than the extent of tumour thrombus [11,23]. An analysis of our institutional experience shows that the development in intraoperative monitoring, surgical technique and prudent use of sophisticated vascular bypass techniques lessens the incidence of profound haemodynamic changes that can lead to death, visceral injury and coagulopathy. These advances have led to decreased mortality, perioperative morbidity and overall improved surgical outcome for patients who present with the most challenging tumours treated in genitourinary oncology. This review of the present series of patients with renal venous tumour extension from RCC, including 191 with IVC tumour thrombus, again shows the importance of primary tumour histological factors as important predictors of prognosis. In addition, patients with thrombus only in the renal vein (level 0) had significantly better survival than those with IVC involvement (levels >0). However, there was no significant difference in survival by thrombus level among patients with IVC thrombus.

Along with improvements in cross-sectional imaging that allow better definition and determination of the extent of intravascular RCC extension, there has been improved intra-and perioperative monitoring of patients [12,13,15–17,20]. As patients are diagnosed with a higher thrombus level our experience indicates that they are subject to more complications, and therefore need more aggressive and careful surgical planning (i.e. the early complication rate for patients with level II thrombus was 14% vs 18% and 30% for those with level III and IV tumour thrombus, respectively).

In planning the resection of IVC tumour thrombus, the recognition of partial vs complete IVC occlusion is important for appropriate planning. Paradoxically, patients with complete occlusion very commonly require no vascular bypass as they have completely collateralized and tolerate proximal clamping of the IVC either sub- or superdiaphragmatically quite well. For example, 10 (36%) of the present patients with level III thrombus had a completely occluded vena cava, but only four required a vascular bypass technique. Indeed, no extracorporeal circulatory bypass was necessary in 71% of patients with level III occlusion. However, of the 20 patients with level IV tumour thrombus, none had complete occlusion and all required a vascular bypass technique, most commonly cardiopulmonary with circulatory arrest, in 13. Although the differences are not significant, patients with complete occlusion tended not to require a bypass as often and had fewer early complications (21% vs 17%, P= 0.577).

Intraoperative TEE is used to monitor the progress of the thrombus during the mobilization phase of the operation. We think that the judicious use of vascular bypass is the primary reason for the decreased morbidity of this procedure. Specifically, the incidence of any early surgical complication decreased from 13% for patients treated in 1970–1989 to 8% for those treated in 1990–2000, a decade when these imaging and surgical techniques were adopted. Operative mortality decreased from 3.8% to 2.0% between the periods, and among patients with IVC tumour thrombus it decreased from 8.1% to 3.8%.

Recently, venous bypass techniques with an extracorporeal membrane oxygenator have been used effectively in 3%, 7% and 30% of patients with level II, III, and IV thrombus, respectively [15]. There was a significant decrease in the duration of surgery and anaesthesia when using a venous rather than a cardiopulmonary bypass. The median cardiopulmonary bypass time of 48 min for level III thrombus and 87 min for level IV thrombus compares favourably with those in other series [3–5,10], at 57, 50, 23.5 and 160 min, respectively. The perioperative complication rate of patients with level IV thrombus who went on a venous bypass was 17%, compared with 31% for those who had a cardiopulmonary bypass and circulatory arrest. We think that avoiding prolonged circulatory arrest, the need for anticoagulation (which is not required for venous bypass), and prolonged anaesthesia for warming, decreased the risk of coagulopathy. Intraoperative TEE allows an early assessment of the tumour thrombus characteristics that predict for a free-floating and easily extracted thrombus either in the superdiaphragmatic vena cava or right atrial chamber. Venous bypass is then considered to avoid hypotension during cavotomy and tumour thrombus extraction. It also allows an unhurried inspection and repair of the IVC. The results from the current study suggest the further use of this technique, if prudent, for level III or IV tumour thrombus.

The vena cava was resected in 10 of 20 patients with level IV thrombus and patched with either Gore-tex or pericardium, compared with only one (4%) with a level III thrombus. Interestingly, the diameter of level III thrombus appeared to be greater as there was complete occlusion in 36%, compared with none in level IV thrombus. Therefore, the need for resection of the vena cava cannot necessarily be predicted beforehand by the level of tumour thrombus, partial or complete occlusion, or diameter of tumour thrombus. Prudent use of vena cava clips or IVC filters is predicated upon the presence of bland thrombus distal to the tumour thrombus. The presence of tumour thrombus can be predicted in most instances by MRI and correlates closely with complete venous occlusion. If all of the blood thrombus cannot be mobilized and extracted, then consideration of a vena cava clip (9%, or 17 of 191) to prevent perioperative embolus would seem to be prudent.

There was a statistically significant increase in the incidence of long-term complications over time, reflecting an increase in long-term proteinuria and chronic renal insufficiency. This is probably a result of the significant increase in patients with level II–IV IVC tumour thrombus over time (Table 3). This increased incidence reflects our referral-based practice, and improvements in surgical techniques and imaging that allow for more aggressive approaches with these tumours, rather than a true stage migration. The 5.8% operative mortality (11 of 191) of patients with IVC tumour thrombus is consistent with those reported in others of 2.7–13%[1–5,8–10]. However, the operative mortality of 2% among all patients treated with tumour thrombus between 1990 and 2000 is significantly less than the 9.3% operative mortality reported by Neves and Zincke [6] in 1986 from our institution.

The prognostic implication of the level of tumour thrombus has been extensively analysed; most series show no significant difference in survival based on the extent of tumour thrombus. However, some have found that tumour thrombus extending to the atrium affects survival more adversely than subdiaphragmatic tumour thrombus [1–10,25]. In previous investigations the 5-year survival rate for patients with venous involvement and no evidence of metastases at surgery was 30–72%, but in those with metastases at surgery it was 0–25%[1–10,25–27]. We showed previously that complete resection of IVC tumour thrombus in the absence of nodal or distant metastases results in a 5-year survival of 50%[6]. The current analysis supports earlier data, in that patients with pN0/pNx, pM0 disease had a 59% 5-year cancer-specific survival, compared with 6% for patients with pN1/pN2, pM1 disease (Fig. 1C). In a recently reported series [28] there was no statistically significant difference in the risk of recurrence for patients with pN0, pM0 disease with renal vein involvement compared with those with pN0, pM0 disease and IVC involvement. In the present 540 patients with RCC and venous extension there was no significant difference in outcome by IVC thrombus level. However, patients with level 0 renal vein thrombus had significantly better survival than those with IVC tumour thrombus. Patients with evidence of positive regional lymph nodes and pM1 disease with renal venous extension had a 5-year cancer-specific survival of 6%, compared with those with pN0/pNX pM1, disease (15%) and those with pN1/pN2, pM0 disease (17%). These data are in agreement with other series [3,6–8], but conversely, others reported no adverse effect on survival in patients with or with no lymph node metastases [4,25]. More recent data [12] confirm a 5-year cancer-specific survival of 33% with lymph node positive disease, vs 59% in the absence of retroperitoneal lymph node dissection. The present data confirm an earlier report by Neves and Zincke [6], wherein pN1 or pN2 diseases associated with pM1 disease portended a poorer prognosis than pN0, pM1 disease.

None of the present patients received adjuvant immunotherapy but the survival according to tumour level appeared to be similar to those reported in the series using cytoreductive surgery and adjuvant immunotherapy [28]. The major difference in these two series is that the latter cohort consisted of only a third of patients with no nodal or metastatic disease, compared with two-thirds of the present patients.

The histological features of the primary tumour determine the biological potential and tumour aggressiveness [24]. Univariate and multivariate analyses show that histological subtype, histological grade, perinephric fat invasion, and nodal and metastatic status are significant prognostic features [11]. In addition, the current analysis showed that histological tumour necrosis and sarcomatoid component were markers for a poor outcome and should be considered as stratification factors in the design of future clinical trials. For instance, the 5-year cancer-specific survival rates for patients with and with no tumour necrosis were 26% and 61%, respectively (P < 0.001); the rates for patients with and with no sarcomatoid component were 3% and 47%, respectively (P < 0.001).

This retrospective analysis is limited by all the inherent biases of reporting one institutional series. However, the wide variety of presentation of these patients, improvements in surgical technique and morbidity, the recent report of a lack of effect of comorbid disease on clinical outcome, and possible improved response to immunotherapy with cytoreductive surgery, precludes the development of prospective trials of surgery vs observation [11,29]. These issues must be considered when comparing these data to other historical large series of patients who have undergone resection with or with no adjuvant immunotherapy. We think that the present large series is important because the patients were treated prospectively without the benefit of adjuvant immunotherapy, and therefore stand as a control to other datasets that report survival after resection and immunotherapy. Realistically, a multi-institutional effort, prospective, randomized and controlled, with patients with extensive RCC, is needed to determine the effect of adjuvant immunotherapy in these patients. Having stated this, our institutional position is that even patients with regional or distant metastatic disease should be considered for resection of these extensive lesions to relieve symptoms, remove an immediate life-threatening aspect of the disease, and possibly reduce the tumour burden to improve the patient's ability to respond to immunotherapy [29]. The judicious use of appropriate bypass techniques, and in particular consideration of venous bypass, reduces the associated problems such as coagulopathy, hepatic dysfunction and renal insufficiency, with an acceptable overall operative mortality of 2% in the most recent decade of surgical experience.


None declared.