Factors associated with surgery in patients with renal cell carcinoma and venous tumor thrombus

Authors


Jared M Whitson, University of California San Francisco, Box 1695, 1600 Divisadero Street, A-633, San Francisco, CA 94143, USA. e-mail: jwhitson@urology.ucsf.edu

Abstract

Study Type – Therapy (case series)

Level of Evidence 4

What’s known on the subject? and What does the study add?

This is the first published report which is an in-depth analysis of factors associated with surgery versus non-operative management in patients with renal cell carcinoma and venous tumour thrombus.

OBJECTIVE

• Venous tumour thrombus is common in patients with renal cell carcinoma (RCC). Although surgical morbidity has decreased with time, nephrectomy with caval thrombectomy remains a high-risk procedure and may not be performed in all patients with this condition. Little is known about the factors influencing the decision to pursue surgery versus conservative management in patients with RCC and venous tumour thrombus.

MATERIALS AND METHODS

• The Surveillance, Epidemiology, and End Results database was used to identify study patients with RCC and venous tumour thrombus.

• Multiple clinical, pathological and sociodemographic variables were assessed.

• Univariable and multivariable logistic regression analysis was performed to identify factors associated with surgery.

RESULTS

• We identified 24 396 patients with RCC, of which 2265 (9.3%) had venous tumour thrombus.

• Distant metastases (odds ratio [OR] 0.1, 95% CI 0.0–0.1), clinical stage T3c (OR 0.3, 95% CI 0.2–0.6), lymph node involvement (OR 0.4, 95% CI 0.2–0.6), being single (OR 0.4, 95% CI 0.3–0.7), and the age categories 61–70 years (OR 0.4, 95% CI 0.2–0.8, P= 0.01), 71–80 years (OR 0.2, 95% CI 0.1–0.3, P < 0.001), and ≥80 years (OR 0.1, 95% CI 0.0–0.1, P < 0.001) were significantly associated with non-surgical management.

CONCLUSIONS

• In this population-based study, over 80% of patients with RCC and venous tumour thrombus underwent surgical management.

• Although age and TNM stage were strongly associated with the decision to undergo surgery, marital status was also associated with treatment choice.

• It is unclear whether marital status affects oncological outcomes or complication rates so the reasons behind this association deserve further investigation.

Abbreviations
OR

odds ratio

SEER

Surveillance Epidemiology and End Results

VTT

venous tumour thrombus

INTRODUCTION

Venous tumour thrombus (VTT) are common in patients with RCC, occurring in approximately 5–10% of all cases [1]. Once considered a disease with minimal treatment options, refinements in surgical techniques in the 1970s resulted in decreased morbidity of caval thrombectomy and led to a transition toward aggressive surgical management [2]. Initial studies found encouraging results in patients who underwent surgical resection, with long-term survival rates similar to patients without VTT [3].

Nonetheless, the presence of VTT is often associated with larger tumours and nodal or metastatic disease [4]. Surgical management was initially reserved for a limited number of patients because it was unclear whether patients with metastases would benefit from this aggressive approach [5]. More recently, however, data have emerged regarding the benefits of cytoreductive nephrectomy in the setting of metastatic RCC [6], expanding the indications for surgical intervention.

As a result, most of today’s patient with RCC and VTT are treated with radical nephrectomy and caval thrombectomy. However, despite the encouraging outcomes in selected patients, this procedure often results in significant morbidity and mortality [7]. Mortality rates as high as 40% have been reported for patients with thrombus extending above the diaphragm [8].

Given the significant risks of caval thrombectomy, not all patients are considered surgical candidates. The availability and efficacy of novel targeted therapy has provided a potential alternative management strategy in those patients with advanced disease. There are few data regarding the natural history and prognosis of RCC with VTT. Case reports of conservatively managed patients have reported a median survival of 3–5 months [3,8] and 2-year mortality rates approaching 95%[9]. These results, however, are probably biased because patients excluded from surgery were more likely to harbour significant comorbidities or widespread metastatic disease.

The risks and benefits of caval thrombectomy versus conservative management must be considered in each individual patient. Unfortunately, there are few guidelines in the literature indicating which patients would benefit from each approach. Additionally, little is known about the factors used by urologists to determine who is a candidate for aggressive surgical management. We performed an analysis of the Surveillance Epidemiology and End Results (SEER) database to identify factors associated with surgical management in patients with RCC and VTT.

MATERIALS AND METHODS

The SEER database was used to identify potential patients. SEER is a large US population-based cancer registry administered by the National Cancer Institute. It collects data on incident cancer cases from 18 individual cancer registries, represents approximately 26% of the US population and mirrors its demographic composition.

Our cohort included patients with RCC and VTT diagnosed after 2004. Patients with renal tumours were initially identified by including those with ‘Site rec B’ listed as kidney. Patient with an ‘ICD-O-3’ of 8120, indicating transitional cell carcinoma, were excluded. The study population was further refined to ensure that only RCC histology was included, based on ‘histologic type ICD-O-3’, including: clear cell (8310), papillary (8050, 8260, 8342), chromophobe (8270, 8290, 8317), collecting duct (8319), medullary (8510), granular (8320), sarcomatoid (8318) cancers and RCC not otherwise specified (8312). The presence of VTT was insured by including only those patients with stage T3b or T3c disease [10]. Paediatric patients (age <18 years) were excluded, as were those with missing information on surgery and with year of diagnosis before 2004. Before 2004, TNM staging within SEER was less accurate, and patients with VTT were coded as having distant disease regardless of the presence of metastases.

Potential predictor variables to be analysed included: age at diagnosis, gender, race/ethnicity, marital status, SEER region, county attribute, previous cancer diagnosis, month of diagnosis, tumour size by T stage category, T3b or T3c, N stage, and M stage. Race/ethnicity was defined as White, Black, Hispanic, and other (comprising Asian, American Indian/Alaska Native and Native Hawaiian/Pacific Islanders). Marital status was initially examined as married, single and other (comprising divorced, separated and widowed). SEER registries were grouped into the geographic locations by the regions Central, Northeast, Southeast and West. The county attribute was defined using the SEER Rural-Urban Continuum code as a rural area (codes 4–9 and 88) or an urban area (codes 0–3). The percentages of people of similar race living within a patient’s county with less than a high school education and living below the poverty line were recorded. The median income for each patient’s county was also examined.

Surgery, as opposed to non-operative management, was the primary outcome measured. ‘Rx Surgery Primary Site 1998+’ was used to identify the type of surgery as follows: no surgery (0), ablation (10–25), partial nephrectomy (26–39), radical nephrectomy (40–89), nephrectomy not otherwise specified (90) and unknown (99). Patients undergoing no surgery or ablation were combined into a no surgery group, whereas patients treated with partial nephrectomy, radical nephrectomy or nephrectomy not otherwise specified were combined into a surgery group.

Univariable logistic regression was used to examine the associations between the predictor and outcome variables. Variables that showed at least a trend (P= 0.10) toward an association with a patient undergoing surgery were included in the multivariable model. The α-value was set at 0.05 and 95% CI were determined. Data were analysed using stata® version 11.0 (StataCorp, College Station, TX, USA).

RESULTS

A total of 2265 patients met the inclusion criteria. As a total of 24 396 patients were diagnosed with RCC during this time period, the incidence of VTT was 9.3%. Not surprisingly, of the patients with VTT, only 25 (1.1%) underwent partial nephrectomy.

The baseline clinical and pathological characteristics of the study cohort are summarized in Table 1. Median patient age was 64 years (range 24–95). The median tumour size was 8 cm (range 0–30). Vena caval thrombus above the diaphragm (stage T3c) was present in 137 (6%) patients. Approximately two-thirds of the study population presented with node positive and/or distant metastatic disease.

Table 1. Clinical and pathologic characteristics of the study cohort
CharacteristicNo. Patients(%)
  • *

    Missing data present in the following no. subjects (%): Tumor Size 104 (5), N Stage 173 (7), M Stage 62 (3).

Age (years)  
 <50241 (11)
 51–60552(24)
 61–70682(30)
 71–80548(24)
 80+242 (11)
Gender  
 Female774(34)
 Male1491(66)
First cancer  
 No272(12)
 Yes1993(88)
Tumor size*  
 <4 cm125(6)
 4–7 cm773(34)
 >7 cm1263(56)
T stage  
 T3b2128(94)
 T3c137(6)
N stage*  
 N01691(75)
 N+401(18)
M stage*  
 M01509(67)
 M+694(31)

Sociodemographic characteristics can be found in Table 2. Most patients were White (72%) and married (66%). Within the SEER regions, there is an over-representation of people living in the western USA (59%). Most people lived within an urban area (88%). The mean percentage of people of the same race living in the study subjects’ counties who had less than a high school education was 19.2% (range 2–71%). The mean percentage of people of the same race living in the study patients’ counties who were living below the poverty line was 11.1% (range 2–61%). The average median income for the counties in which the study patients lived was $46 800 (range $18 000–79 900).

Table 2. Sociodemographic characteristics of the study cohort
CharacteristicNo./Mean(%/range)
  • *

    Missing data present in the following no. subjects (%): Race 11 (<1), Marital Status 72 (3), Rural-Urban 13 (<1).

  • †Includes American Indian, Alaska Native, Native Hawaiian, and Pacific Islander.

  • ‡Includes separated, divorced, widowed.

Race*  
 White1629(72)
 Black150(7)
 Hispanic318(14)
 Asian/Other157(7)
Marital status*  
 Married1447(66)
 Single272(12)
 Other474(21)
Region  
 Central272(12)
 Northeast326(14)
 Southeast324(14)
 West1343(59)
Rural-Urban*  
 Rural270(12)
 Urban1982(88)
<H.S. education (%)19.2(2–71)
<Poverty line (%)11.1(2–61)
Median income ($1k)46.8(18–79)

Of the entire study cohort, 1875 (82.8%) patients underwent surgery whereas 390 (17.2%) were managed conservatively. The proportion of patients undergoing surgery did not change during the study period (range 82–83%, data not shown).

The results of the univariable logistic regression for factors associated with surgery are summarized in Table 3. Compared with patients <50 years, those between 61 and 70 years [odds ratio (OR) 0.6, P= 0.05], between 71 and 80 years (OR 0.3, P < 0.001) and ≥80 years (OR 0.2, P < 0.001) all had lower odds of surgery. Men had greater odds of surgery than women (OR 1.4, P= 0.005). Race was also associated with surgical management as compared with White patients, both Black (OR 0.6, P= 0.02) and Hispanic (OR 0.7, P= 0.05) patients had lower odds of surgery. Separated, divorced or widowed patients had significantly lower odds of surgery than married patients (OR 0.5, P < 0.001), and single patients showed a similar association (OR 0.7, P= 0.04). Patients living in a county with a higher percentage of people below the poverty line had lower odds of surgery (OR 0.8, P= 0.003). Finally, tumour size >7 cm (OR 0.3, P= 0.004), VTT above the diaphragm (OR 0.3, P < 0.001), node positive disease (OR 0.2, P < 0.001) or metastatic disease (OR 0.1, P < 0.001) were all associated with non-surgical management.

Table 3. Univariable analyses of predictors of surgery
Variable (referrent)OR95% CI P-value
  • *

    Statistically significant, P < 0.05.

  • †Confidence interval appears to include 1.0 because of rounding.

  • ‡Includes American Indian, Alaska Native, Native Hawaiian, and Pacific Islander.

  • §

    §Includes separated, divorced, widowed.

  • ¶Continuous variable.

Age (<50)  <0.0001*
 51–600.70.4–1.10.11
 61–700.60.4–1.00.05*
 71–800.30.2–0.5<0.001*
 80+0.20.1–0.3<0.001*
Male (female)1.41.1–1.70.005*
Race (white)  0.01*
 Black0.60.4–0.90.02*
 Hispanic0.70.5–1.00.05*
 Asian/Other1.40.9–2.30.19
Marital status (married)  <0.0001*
 Single0.70.5–1.00.04*
 Other§0.50.4–0.7<0.001*
First cancer (other cancers)1.20.9–1.70.18
Region (central)  0.11
 Northeast1.30.9–1.80.12
 Southeast1.41.0–2.00.05*
 West1.00.7–1.40.95
Urban (rural)1.10.8–1.60.47
<H.S. education (per 10% inc)0.90.9–1.00.09
<Poverty line (per 10% inc)0.80.7–0.90.003*
Median income (per $10k inc)1.10.9–1.20.13
Tumor size (<4)  <0.0001*
 4–70.60.3–1.20.13
 >70.30.2–0.70.004*
T3c (T3b)0.30.2–0.4<0.001*
N+ (N0)0.20.1–0.3<0.001*
M+ (M0)0.10.0–0.1<0.001*

In a preliminary multivariable model adjusting for age, TNM stage, gender, tumour size, race and marital status, Black patients had a significantly lower odds of surgery than White patients (OR 0.4, P= 0.006). However, after inclusion of the county-wide poverty level in the full multivariable model, Black versus White was no longer associated with treatment strategy (OR 0.6, P= 0.26).

The results of the full multivariable logistic regression can be found in Table 4. Similar to the univariable analysis, the odds of surgery were lower in patients with VTT above the diaphragm (OR 0.3, P < 0.001), node positive disease (OR 0.4, P < 0.001), distant metastases (OR 0.1, P < 0.001) and in patients between 61 and 70 years (OR 0.4, P= 0.01), 71 and 80 years (OR 0.2, P < 0.001) and ≥80 years (OR 0.1, P < 0.001). Married patients still had greater odds of surgery than single, separated, divorced or widowed patients, although in the multivariable model, only the OR for the single patients reached statistical significance (OR 0.4, P < 0.001).

Table 4. Multivariable analysis of predictors of surgery
Variable (referrent)OR95% CI P-value
  • *

    Statistically significant, P < 0.05.

  • †Includes American Indian, Alaska Native, Native Hawaiian, and Pacific Islander.

  • ‡Includes separated, divorced, widowed.

  • §

    §Continuous variable.

Age (<50)  <0.0001*
 51–600.70.3–1.40.27
 61–700.40.2–0.80.01*
 71–800.20.1–0.3<0.001*
 80+0.10.0–0.1<0.001*
Male (female)1.10.8–1.60.57
Race (white)  0.32
 Black0.60.2–1.40.26
 Hispanic0.90.4–1.70.65
 Asian/Other1.60.8–3.30.20
Marital status (married)  0.003*
 Single0.40.3–0.70.001*
 Other0.70.5–1.00.08
<Poverty line (per 10% inc)§0.80.6–1.20.26
Tumor size (<4 cm)  0.18
 4–70.60.2–1.60.28
 >70.40.2–1.30.12
T3c (T3b)0.30.2–0.6<0.001*
N+ (N0)0.40.2–0.6<0.001*
M+ (M0)0.10.0–0.1<0.001*

Because patients undergoing partial nephrectomy might represent a unique subset of the overall cohort, we performed a separate logistic regression without them included. There were no significant differences observed between these models and the original ones (data not shown).

DISCUSSION

Encouraging outcomes have been reported for selected patients with RCC and VTT treated with radical nephrectomy and caval thrombectomy. Accordingly, most such patients are treated surgically. In our cohort, 82.8% of patients were treated with nephrectomy, although this leaves a significant percentage that was managed conservatively. We examined a number of tumour-specific, sociodemographic and geographic variables to identify those factors associated with surgical management.

Our data indicate that tumour-specific factors were most strongly associated with surgical management. Specifically, patients with advanced T, N or M stage had the lowest odds of undergoing nephrectomy. Of these, metastatic disease was most related with non-aggressive management. This would suggest that a strong consideration of the risk-benefit profile of surgery versus conservative management is made before undergoing radical nephrectomy with VTT. Although surgical complication rates appear to be similar in patients with and without metastases [11], five-year survival in patients with metastases and VTT is estimated at only 17%[12] compared with the 60% survival rate expected in those without metastases [13].

Patients with evidence of nodal involvement also had lower odds of surgery. Although these patients may harbour more advanced disease, several reports in the literature have shown that patients without distant metastases can be rendered disease free by radical nephrectomy and lymphadenectomy [14]. The literature in patients with concomitant VTT is sparse, if these results could be generalized to patients with nodal involvement and VTT one could argue that patients with nodal disease should not be excluded from aggressive surgical management.

More cephalad extent of thrombus was also strongly associated with conservative management. This probably reflects the increased technical difficulty and complication rates of caval thrombectomy with more extensive VTT. One large study reported a step-wise increase in early complication rates from 12% in those with renal vein thrombus to 47% in those with right atrial thrombus [15]. Additionally, surgical blood loss is significantly greater in patients with a greater extent of VTT [15]. Finally, surgery involving cardiopulmonary [16] or veno-venous bypass [17] is probably more intimidating from both the patient’s and the physician’s perspective, and may dissuade them from pursuing surgical management. In the future, down-staging with tyrosine kinase inhibitors may have a role [18].

Tumour size was the one tumour-specific factor that was not found to be related to surgical management in our study. This seems appropriate, as although larger tumours may result in more technically challenging surgery, tumour size is not an independent predictor of long-term oncological outcome [13,19]. This finding is contrast with studies of cytoreductive nephrectomy, in which patients with larger tumours were more likely to undergo surgery [20]. However, the goal of cytoreductive nephrectomy is to debulk tumour burden and therefore surgery for small primary tumours may not be advantageous.

Several demographic variables were also associated with the odds of undergoing surgery. Older patients had significantly lower odds of undergoing surgery. This may reflect more numerous comorbidities in elderly patients and the increased risk of surgery associated with these comorbidities. Not only is there an association between advanced age and more extensive comorbidities, but one study also found that older patients had considerably worse cancer outcomes after high-risk cancer operations [21]. Furthermore, another study showed that even when controlling for comorbidities, advanced age is independently associated with worse short-term outcomes after major oncological resections [22].

In univariable analysis, race was found to be associated with the decision to pursue operative versus conservative management. Similar results have been reported in several earlier studies [20,23–25] and investigators have suggested that this may be related to more extensive comorbidities in Black patients [23]. However, these studies either did not account for education or poverty level, or adjusted for county-wide income level without regard to race. Although SEER data do not contain comorbidity data and so we could not directly assess this hypothesis, we found that once we adjusted for a county-level poverty variable, an association between race and surgery was no longer observed. This suggests that financial limitations in minorities may restrict their access to healthcare [26]. One report showed that Black patients may be more reluctant than White patients to undergo surgery, a question that we were unable to address with these data [27].

Finally, we found that single patients had lower odds of surgery than married patients. A similar result was recently reported for patients with metastatic RCC without VTT [24]. It has been hypothesized that these differences may be the result of pressure from the spouse to pursue more aggressive treatment, or of the physician’s beliefs that postoperative outcomes are superior in patients with strong support networks. Interestingly, marriage has also been shown to impart a survival benefit in patients with a number of urological malignancies, including renal [24], bladder [28] and prostate [29] carcinomas.

There are several limitations to our study. First, coding for surgery takes place within the 4 months after diagnosis. It is possible that the interval from diagnosis to surgical management is longer in elderly or single patients or in those with more advanced disease, perhaps because of decreased performance status. This may lead to differential misclassification of the outcome, biasing the results away from the null. Second, there could be residual confounding by socioeconomic status because we used race-specific county-level data rather than individual-level data. In the future, using the linked SEER-Medicare database could allow for adjustment for these confounders. Third, in determining the risk-benefit ratio of nephrectomy with caval thrombectomy, providers are almost certainly influenced by comorbidities. As the SEER database does not contain detailed information regarding comorbidities, their impact on surgical decision-making could not be assessed. Finally, the outcomes of the patients treated with surgery versus conservative management in this study are not known. For these reasons, although our study offers insight into factors influencing the decision to pursue surgery, it does not indicate which patients are appropriate for surgical versus expectant management.

In this population-based study, over 80% of patients with RCC and VTT underwent surgical management. Although age and TNM stage were strongly associated with the decision to undergo surgery, marital status was also associated with treatment choice. As it is unclear whether marital status affects oncological outcomes or complication rates, the reasons behind this association deserve further investigation.

ACKNOWLEDGMENTS

Jared M Whitson is supported by funding from the American Urologic Association Research Scholars Program.

CONFLICT OF INTEREST

None declared.

Ancillary