Predictors of Survival in Patients Undergoing Surgery for Renal Cell Carcinoma and Inferior Vena Cava Tumor Thrombus

Abstract

Introduction:

Surgical resection of renal cell carcinoma (RCC) with inferior vena cava (IVC) thrombus is a complex procedure with significant morbidity. Patient selection is critical to determining whether the benefits of the procedure outweigh the risks. In this study, we identified and stratified the risk factors that were associated with overall survival (OS) and recurrence-free survival (RFS) in patients undergoing surgical resection of RCC with IVC thrombus.

Methods:

We identified all patients with RCC with IVC tumor thrombus (stages cT3b and cT3c) who had undergone radical nephrectomy with tumor thrombectomy between December 1, 1993 and June 30, 2009. Kaplan-Meier method was used to estimate OS and RFS. Cox proportional hazards models were used to determine the association between risk factors and OS. Patients were stratified into 3 groups based on the number of risk factors present at diagnosis.

Results:

Two hundred twenty-four patients were included in the study. A total of 45.3% of patients had metastasis at presentation, 84.5% had cT3b, and 90.2% had clear cell RCC. cT3c, cN1, and cM1 were significantly associated with the risk of death. Group 1 patients (0 risk factors) had a median OS duration of 77.6 months (95% CI 50.5-90.4), group 2 (1 risk factor) 26.0 months (95% CI 19.5-35.2), and group 3 (≥2 risk factors) 8.9 months (95% CI 5.2-12.9; p<0.001).

Conclusions:

Stratification of patients with RCC and IVC thrombus by risk factors allowed us to predict survival duration. In patients with ≥2 risk factors, new treatment strategies with preoperative systemic therapy may improve survival.

Micro-abstract:

We assessed preoperative factors associated with overall survival in patients with renal cell carcinoma with IVC thrombus (cT2b, cT3c). We confirmed that cT3c, cN1, and cM1 were independently associated with poor OS. In patients with ≥2 risk factors, the median overall survival was 8.9 months. Stratification of patients before surgery allows us to determine whether the benefits of the procedure outweigh the risks.

INTRODUCTION

Renal cell carcinoma (RCC) with inferior vena cava (IVC) thrombus (American Joint Committee on Cancer stages cT3b and cT3c) represents 4% to 10% of all RCC cases¹. Currently, the only curative treatment available is radical nephrectomy with IVC thrombectomy, a complex procedure that is associated with significant morbidity². Because of the low incidence and complexity of the procedure, most surgeries require a multidisciplinary team and are performed at high-volume centers^3-5.

Patient selection is critical to determining whether the benefits of radical nephrectomy with IVC thrombectomy outweigh the risks of the procedure. Multiple efforts have been made to identify the clinical and pathologic risk factors that are associated with recurrence and survival^6-11. In this study, we identified and stratified the preoperative risk factors that were associated with recurrence and survival in patients with RCC with IVC tumor thrombus.

PATIENTS AND METHODS

After obtaining Institutional Review Board approval, we retrospectively reviewed The University of Texas MD Anderson Cancer Center (Houston, Texas) nephrectomy registry and identified 265 patients with RCC with IVC tumor thrombus (stages cT3b and cT3c) who had undergone radical nephrectomy with tumor thrombectomy between December 1, 1993 and June 30, 2009. ¹

We collected patients’ demographic and clinical features: age, sex, race, performance status, and clinical stage. The pathologic features collected included histologic type, disease stage per the 2017 American Joint Committee on Cancer (8th edition) TNM standard (cT stage, cN stage, and cM stage), and the presence of sarcomatoid differentiation. Data were abstracted via a direct chart review by the authors. Patients with incomplete data were excluded.

Patient characteristics were summarized using the frequency (%) and median (interquartile range) for categorical and continuous variables. Categorical and continuous data were compared using the Fisher’s exact test and Kruskal-Wallis test, respectively. The overall survival (OS) duration was defined as the time between the date of surgery and death and was censored at the last follow-up date for patients who were still alive.

The Kaplan-Meier method was used to estimate OS and recurrence-free survival (RFS) durations from the date of radical nephrectomy and IVC thrombectomy. We compared time-to-event endpoints by subgroups using the log-rank test. X-tile software was used to determine a clinically relevant cut-point for age for OS¹². A multivariate Cox proportional hazards model was then fitted for OS and RFS by including all statistically significant covariates from univariate Cox models. In addition, a landmark analysis of OS by recurrence location, starting on the recurrence day, was estimated using the Kaplan-Meier method. Patients with metastatic disease at presentation were excluded from the landmark and recurrence analyses.

The full multivariate model included all the variables from the univariate analyses with p<0.25 (Table 1). The variables with missing values ≥25%, such as operative time and performance status, were excluded from the full model; we also excluded variables with collinearity issues, such as cytoreductive nephrectomy. Backward elimination methods were used to identify the final reduced model. The backward elimination method fits the full model on all explanatory variables while identifying the least significant variable as “insignificant” (p>0.05); it removes this variable and re-estimates the model without it. These steps are repeated until the model only includes “significant” variables (p<0.05).

Table 1.

Characteristics of the study population (n=265)

Characteristic	Result
Median age at surgery, years (IQR)	60.3 (54.2-70.2)
Median hemoglobin, g/dl (IQR)	11.90 (10.3-13.4)
Median corrected Ca, mg/dl (IQR)	9.30 (9.1-9.8)
Median LDH (IU/L) (IQR)	522.0 (422-693)
Median creatinine, mg/dl (IQR)	1.30 (1.0-1.5)
Median length of stay, days (IQR)	8.00 (6.0-13.0)
Sex, n (%)
Male	175 (66.0)
Female	90 (34.0)
Race and ethnicity, n (%)
White	197 (74.3)
Hispanic	55 (20.8)
Black	10 (3.8)
Asian	3 (1.1)
Clinical T stage, n (%)
3b	224 (84.5)
3c	41 (15.5)
Clinical N1	46 (17.4)
Clinical M1	120 (45.3)
Pre-surgery embolization, n (%)	50 (18.9)
Histologic type, n (%)
Clear cell	238 (90.2)
Non-clear cell	27 (9.9)
Grade, n (%)
0-2	25 (9.5)
3	118 (45.0)
4	119 (45.4)
Sarcomatoid, n (%)	29 (11.0)
Rhabdoid, n (%)	13 (5.0)
Congestive heart failure, n (%)	2 (0.8)
Cytoreductive nephrectomy, n (%)	120 (45.3)
Side, n (%)
Right	80 (30.2)
Left	184 (69.4)
Both	1 (0.4)
Hypertension, n (%)	135 (50.9)
Diabetes mellitus, n (%)	51 (19.3)
Coronary artery disease, n (%)	26 (9.8)
Peripheral vascular disease, n (%)	2 (0.8)
Hematuria, n (%)
Microhematuria	13 (4.9)
Gross hematuria	106 (40.2)
Flank pain, n (%)	78 (29.6)
Weight loss >10 lbs, n (%)	81 (30.7)
Malaise weakness fatigue, n (%)	81 (30.7)
Bilateral leg edema, n (%)	40 (15.2)
New-onset congestive heart failure, n (%)	13 (4.9)

Risk Score

OS and RFS risk scores were obtained by rounding each result to the nearest integer, for OS using the group with the lowest risk of dying, and for RFS using the group with lowest risk of recurrence as the reference. For OS, the presence of cT3c, cN1, and cM1 disease each equaled 1 point. cT3b, cN0, and cM0 equaled 0. For RFS, cN1, non-clear cell histologic type, and weight loss >10 lbs were each are worth 1 point, while cN0, clear cell histologic type, and weight loss <10 lbs were 0 points. Patients were stratified into 3 groups based on the number of risk factors: group 1, no risk factors; group 2, 1 risk factor; and group 3, ≥2 risk factors.

Nomogram

A nomogram was developed on the basis of the fitted Cox proportional hazards model for OS and assessed for calibration and discrimination. Calibration was assessed graphically using plots of the predicted probability of OS at 1 year versus the actual proportion of patients who had not died at 1 year.

We assessed discrimination by calculating the area under the receiver operating characteristics curve for censored data. To adjust for the bias that is associated with evaluating a nomogram’s performance in the same group of patients used to build the nomogram, we repeated the calibration assessment for 500 bootstrapped samples. The bootstrapped samples’ results estimate the nomogram performance that could be expected in a separate but similar patient population. Patients were divided into 5 groups that contained a mean of 50 patients based on the predicted probability of OS. The closer the blue line is to the black line, the better the calibration.

A statistical analysis was performed using Stata/SE version 16.0 statistical software (Stata Corp. LP, College Station, TX) and R software.

RESULTS

Of the 265 patients who we identified with RCC with IVC tumor thrombus, 224 had complete data and were included in the study. Their median age was 60.3 years (IQR, 54.2-70.2 years); 66.0% were male; 84.5% had cT3b, 17.4% had cN1, and 45.3% had cM1; and 90.2% had clear cell RCC. Sarcomatoid features were present in 11.0% of patients (table 1). The median follow-up duration for all patients was 26.9 months (IQR, 10.1-64.7 months).

Overall Survival

At last follow-up, 223 patients had died, including 159 who had died of RCC. In the univariate analysis, hemoglobin, cT3b, cN1, cM1, and hypertension were all associated with OS (all p<0.05). In the multivariable analysis, cT3c (OR 1.83, 95% CI 1.21-2.76, p=0.004), cN1 (OR 2.65, 95% CI 1.83-3.84, p<0.001), and cM1 (OR 2.66, 1.96-3.61, p<0.001) were all independently associated with poor OS (Table 2).

Table 2.

Univariate and multivariable analysis for overall survival

	Univariate Analysis of OS		Multivariable Analysis of OS
			Full Model		Reduced Model
Characteristic	HR (95% CI for HR)	p- value	OR (95% Cl for OR)	p- value	OR (95% CI for OR)	p- value
Age ≥50 (X-tile)	1.42 (9.95-2.12)	0.091
Age ≥60 (median)	1.1 (0.85-2.12)	0.469
Age ≥62 (mean)	1.11 (0.85-1.44)	0.44
Hgb
Below normal	1	.
Normal	0.69 (0.5-0.94)	0.018	0.8 (0.58-1.11)	0.175
Corrected Ca++
Normal	1	.
Above normal	1.13 (0.75-1.7)	0.555
LDH
Normal	1	.
Above normal	1.31 (0.94-1.83)	0.114
Sex (male vs. female)	0.92 (0.7-1.22)	0.572
Race
Non-Hispanic white	1	.
Hispanic	0.94 (0.68-1.32)	0.737
Black	1.28 (0.65-2.52)	0.473
Asian	2.55 (0.81-8.06)	0.11
cT stage (T3c vs. T3b)	1.59 (1.11-2.28)	0.012	1.81 (1.18-2.79)	0.007	1.83 (1.21-2.76)	0.004
cN1 stage	3.08 (2.2-4.31)	<0.001	2.61 (1.79-3.8)	<0.001	2.65 (1.83-3.84)	<0.001
cM1 stage	2.51 (1.91-3.3)	<0.001	2.67 (1.96-3.65)	<0.001	2.66 (1.96-3.61)	<0.001
Pre-surgical embolization Side	0.9 (0.65-1.26)	0.55
Left	1	.
Right	1.23 (0.92-1.63)	0.164
Hypertension	0.75 (0.58-0.98)	0.033	0.98 (0.72-1.33)
Diabetes mellitus	1.09 (0.79-1.51)	0.599
Coronary artery disease	1.26 (0.83-.193)	0.28
Peripheral vascular disease	1.16 (0.29-4.66)	0.838
Microhematuria	0.89 (0.48-1.64)	0.701
Gross hematuria	0.89 (0.68-1.17)	0.405
Flank pain	0.98 (0.73-1.31)	0.868
Weight loss >10 lbs	1.22 (0.92-1.63)	0.169
Malaise weakness fatigue	1.12 (0.84-1.5)	0.427
New-onset congestive heart failure	1.04 (0.56-1.9)	0.911

The risk score was calculated using the variables cT3c, cN1, and cM1. Patients in group 1 (0 risk factors) had a median OS duration of 77.6 months (95% CI 50.5-90.4), group 2 (1 risk factor) 26.0 months (95% CI 19.5-35.2), and group 3 (≥2 risk factors) 8.9 months (95% CI 5.2-12.9; p<0.001) (Figure 1).

Figure 1.

Kaplan-Meier curves of overall survival in patients with renal cell carcinoma and inferior vena cava thrombus, based on the preoperative score.

A nomogram was constructed using cT stage, cN stage, and cM stage to show the probability of OS at 1, 3, and 5 years (Figure 2). Figure 3 shows the plot of the predicted probability of OS at 1 year versus the actual proportion of patients who survived for 1 year without experiencing an event. The black line represents a nomogram with perfect calibration, or one in which the predicted probability is equal to the observed probability. The blue line represents the performance of our nomogram. The area under the receiver operating characteristics curve for this nomogram was 0.65.

Figure 2.

Nomogram of overall survival at 1 year, 3 years, and 5 years on the basis of the fitted Cox model.

Figure 3.

Calibration of the nomogram.

Disease Recurrence

Eighty-eight patients had recurrent disease. The 4 most common recurrence locations were the lungs (47), liver (19), bones (14), and renal fossa (10). In the univariate analysis, cN1, weight loss >10 lbs, non-clear cell histologic type, and sarcomatoid features were all associated with RFS (all p<0.05). In the multivariable analysis, cN1 (HR 2.63, 95% CI 1.51-4.6, p=0.01), weight loss >10 lbs (HR 1.76, 95% CI 1.18-2.65, p=0.006), and non-clear cell histologic type (HR 2.08, 95% CI 1.17-3.69, p=0.01) were all independently associated with poor RFS (Table 3).

Table 3.

Univariate and multivariable analysis of recurrence-free survival

	Univariate Analysis for RFS		Multivariable Analysis for RFS
Characteristic	HR (95% CI for HR)	p- value	HR (95% CI for HR)	p- value
Age ≥50 (X-tile)	1.33 (0.79-2.26)	0.284
Age ≥60 (median)	0.91 (0.64-1.31)	0.617
Age ≥62 (mean)	0.95 (0.64-1.36)	0.78
Hgb
Below normal	1	.
Normal	0.67 (0.44-1.02)	0.065
Corrected Ca++
Normal	1	.
Above normal	1.25 (0.72-2.17)	0.423
LDH
Normal	1	.
Above normal	1.27 (0.8-2.02)	0.318
EGFR
<45	1	.
≥45	0.76 (0.35-1.62)	0.473
Sex
Male	1	.
Female	0.92 (0.64-1.34)	0.678
Race
White	1	.
Hispanic	0.91 (0.59-1.42)	0.686
Black	1 (0.43-2.33)	0.997
T stage
3b	1	.
3c	1.42 (0.87-2.3)	0.157
Clinical N1	2.67 (1.54-4.63)	<0.001	2.63 (1.51-4.6)	0.01
Pre-surgery embolize	0.75 (0.46-1.2)	0.226
Side
Left	1	.
Right	1.06 (0.71-1.6)	0.768
Hypertension	0.78 (0.54-1.12)	0.172
Diabetes mellitus	0.77 (0.49-1.22)	0.267
Coronary artery disease	1.27 (0.76-2.13)	0.362
Peripheral vascular disease	2.67 (0.65-11)	0.175
Microhematuria	1.14 (0.46-2.83)	0.776
Gross hematuria	0.83 (0.56-1.21)	0.328
Flank pain	0.92 (0.63-1.35)	0.67
Weight loss >10 lbs	2.02 (1.36-2.99)	<0.001	1.76 (1.18-2.65)	0.006
Malaise, weakness, fatigue	1.22 (0.83-1.8)	0.3
New-onset congestive heart failure	1.75	0.134
Non-clear cell histologic type	2.29 (1.3-4.04)	0.004	2.08 (1.17-3.69)	0.01
Grade
0-2	1	.
3	0.8 (0.44-1.46)	0.468
4	1.42 (0.77-2.64)	0.263
Sarcomatoid	2.43 (1.38-4.27)	0.002	1.76 (0.98-3.15)	0.054
Rhabdoid	1.43 (0.45-4.53)	0.547

For the RFS risk score, the variables included were cN1, weight loss >10 lbs, and non-clear cell histologic type. Patients were stratified into 3 groups on the basis of their number of risk factors: group 1 (0 risk factors) had a median RFS duration of 37.5 months (95% CI 20-47.8 months), group 2 (1 risk factor) 12.2 months (95% CI 8.38-21.5), and group 3 (≥2 risk factors) 6 months (95% CI 3.41-12.7; p<0.001) (Figure 4).

Figure 4.

Kaplan-Meier curves of recurrence-free survival in patients with renal cell carcinoma and inferior vena cava thrombus, stratified by risk score.

In the landmark analysis of OS by recurrence site, patients with disease recurrence in the renal fossa had a median OS duration of 8 months (95% CI 1-24 months) versus 28 months (95% CI 16-45 months) in those without recurrence in the renal fossa (p=0.01). Patients with liver metastasis had a median OS duration of 11 months (95% CI 7-21 months) versus 30 months (95% CI 19-47 months) in those without liver metastasis (p=0.008); pancreatic metastasis was also associated with a longer OS duration (p=0.005) (Figure 5).

Figure 5.

Kaplan-Meier curves of overall survival in patients with renal cell carcinoma and inferior vena cava thrombus, stratified by recurrence site.

DISCUSSION

Radical nephrectomy with tumor thrombectomy is a morbid procedure, with a 30-day mortality rate between 1.5% and 10%¹³. Patient selection is critical to determining whether the benefit of the procedure outweighs the risk. Our data indicate that preoperative risk factors can be used to stratify patients to predict survival and recurrence.

Multiple studies, including a meta-analysis, have identified thrombus level, tumor size, sarcomatoid differentiation, Fuhrman grade, tumor necrosis, perinephric fat invasion, cN1, and metastasis at presentation as independent predictors of OS^{14, 15}. Our findings are consistent with those of these analyses. To improve patient selection and pre-surgical counseling, our model only included preoperative risk factors and excluded pathologic characteristics. In addition, as not all patients had metastatic disease at presentation, risk scores such as IMDC or MSKCC scores were not applicable.

Patient stratification allows surgeons to select patients objectively and mitigate potential bias. The survival model shows that the presence of ≥2 risk factors is associated with a median OS duration of 8.9 months, making the role of surgery in this group uncertain. Fortunately, in this cohort, patients with ≥2 risk factors only represent 16.5% of the study population, demonstrating that surgery results in a significant improvement in OS duration in the majority of patients with RCC and IVC tumor thrombus.

There is limited evidence regarding the benefit of neoadjuvant or preoperative systemic therapy in patients with RCC with IVC^{16, 17}. In metastatic RCC, immunotherapy, alone or in combination with targeted molecular therapy, has led to significant improvements in OS duration^{18, 19}. A prospective trial of adjuvant sorafenib or sunitinib in patients with non-metastatic RCC and IVC tumor thrombus showed no survival benefit¹⁶. In a small retrospective analysis of patients with metastatic RCC and IVC tumor thrombus, the combination of surgery and targeted therapy had a similar OS duration to that of targeted therapy alone¹⁷. Further prospective trials are needed to delineate the role of surgery in this group of patients with ≥2 risk factors.

The disease recurrence location differs between different histologic subtypes of RCC²⁰. In patients with metastatic clear cell RCC, the most common metastatic sites are the lungs, lymph nodes, bones, and liver. Metastasis to the liver is consistently associated with poor survival. Metastasis to endocrine organs is less common but is associated with improved survival^18-20. These findings are consistent with those in our cohort, in which liver metastasis was associated with poor survival, and pancreatic metastasis had indolent characteristics.

The presence of non-clear cell histologic type and cN1 is consistently associated with disease recurrence in patients with no evidence of metastasis at presentation²¹. Fifty percent of these patients have evidence of recurrence at 6 months and may benefit from clinical trials of new systemic treatments in the neoadjuvant or adjuvant setting. In addition, sarcomatoid histology has been consistently shown to be associated with worse RFS and OS²². In our study, sarcomatoid histology also emerged as a risk factor on multivariable analysis. Immune check point inhibitors (ICI) have been shown to have significant activity in patients with metastatic RCC and sarcomatoid features. A recent meta-analysis of four randomized trials showed that ICI-based systemic therapy combinations in patients with sarcomatoid RCC improved PFS, OS and objective response rate relative to targeted therapies²². In the subset of patients presenting with RCC with IVC tumor thrombus and sarcomatoid histology, the use of these agents in the neoadjuvant or adjuvant setting has the potential to improve outcomes.

The retrospective nature of our study limits our findings. Although we identified risk factors that were associated with a poor prognosis, other factors that are associated with poor prognosis in patients with metastatic RCC and IVC thrombus (systemic symptoms, hemoglobin, and a neutrophils to lymphocytes ratio >4) must be taken into consideration¹⁰. When developing our model, certain potentially prognostic clinical variables, such as performance status, had to be excluded if the data was missing in most patients.

It is possible that the outcomes of our study cannot be extrapolated to the community since MD Anderson is a high-volume center with significant experience with this type of intervention. We believe that surgeons who are experienced with this disease play a significant role in patients’ outcomes and that these patients should be referred to tertiary care centers.

Conclusions

In patients with RCC and IVC tumor thrombus, the preoperative risk factors cT3c, cN1, and cM1 were independent predictors of poor survival. Stratification of patients with RCC and IVC thrombus by risk factors allows us to predict survival duration. In patients with ≥2 risk factors, the benefit of upfront nephrectomy and IVC thrombectomy is uncertain. Our preoperative nomogram allows us to improve patient selection for surgery and to design new trials of preoperative systemic therapy to improve outcomes in these patients.

Clinical Practice Points:

• Preoperative risk factors allowed us to predict survival in patients with renal cell carcinoma with inferior vena cava thrombus.

• In patients with ≥2 risk factors, preoperative systemic therapy may improve survival.

• After surgery, the recurrence location is associated with different overall survival, with some recurrences having an indolent behavior.

Abbreviations:

RCC

Renal cell carcinoma

IVC

Inferior vena cava

OS

Overall survival

RFS

Recurrence-free survival

ICI

Immune Checkpoint inhibitor