EURURO Vol. 71 No. 4

who underwent LND and applying the parameter estimates from this

model to the entire cohort as described in the Supplementary data.

Given heterogeneity in the extent of LND, we conducted a sensitivity

analysis to examine the association of an extended LND, defined as

removal of at least 13 LNs

[17] ,

with oncologic outcomes. PSs for receipt of

extended LND (vs no LND or LND with

13 LNs removed) were obtained

using a logistic regression model. The association of extended LND with

oncologic outcomes was evaluated using adjustment for PS quintile and

IPWs as described above, both for the overall cohort and among cN1

patients.

To further assess the relationship of the extent of LND with oncologic

outcomes, we conducted a secondary analysis to evaluate the association

of the number of LNs removed with the development of distant

metastases, CSM, and ACM among the 606 patients who underwent LND,

of whom 590 had data regarding the number of LNs removed. The

number of LNs removed was modeled as a continuous variable, and Cox

models were adjusted for covariates as described in the Supplementary

data. Univariable subset analyses were performed stratified by the

combination of cN and pN status.

Statistical analyses were performed using SAS version 9.3 (SAS

Institute, Cary, NC, USA) and R version 3.1.1 (R Foundation for Statistical

Computing, Vienna, Austria). All tests were two-sided and

values

0.05

were considered statistically significant.

Results

Overall, 1797 patients underwent RN for M0 RCC, including

606 (34%) with LND. A total of 111 (6.2%) patients were pN1.

The median number of LNs removed was six (IQR 2–12).

Clinicopathologic and radiographic features, stratified by

performance of LND, are summarized in

Table 1 .

After PS

adjustment, these features were well balanced between

patients with and without LND. Specifically, there were no

statistically significant differences in clinicopathologic or

radiographic features after reweighting the cohort by IPWs

( Table 1 )

or within each quintile of PS (data not shown), the

370 matched pairs were well balanced (Supplementary

Table 1), and no feature was associated with receipt of LND

after PS adjustment (data not shown). The median follow-

up among survivors was 10.6 (IQR 6.9–15.0) yr. During this

time, a total of 576 patients develop distant metastases,

472 died from RCC, and 1059 died overall.

In the overall cohort, performance of LND was not

significantly associated with CSM (adjusted for PS quintile:

HR 1.15, 95% CI 0.93–1.43,

= 0.21; stratified by PS quintile:

HR 1.14, 95% CI 0.92–1.41,

= 0.23; IPW: HR 1.03, 95% CI

0.84–1.26,

= 0.78; matched pairs: HR 1.25, 95% CI 0.90–

1.73,

= 0.19) or ACM (adjusted for PS quintile: HR 1.08,

95% CI 0.93–1.27,

= 0.32; stratified by PS quintile: HR 1.08,

95% CI 0.92–1.26,

= 0.36; IPW: HR 1.00, 95% CI 0.87–1.14,

= 0.94; matched pairs: HR 1.09, 95% CI 0.86–1.39,

= 0.46)

using any of the PS techniques

( Table 2

). For development of

metastases, two of the PS techniques suggested a modest

increased risk with LND (adjusted for PS quintile: HR 1.25,

95% CI 1.02–1.52,

= 0.03; stratified by PS quintile: HR 1.24,

95% CI 1.03–1.51,

= 0.03), while the other two PS

techniques revealed no statistically significant association

(IPW: HR 1.13, 95% CI 0.95–1.36,

= 0.18; matched pairs:

HR 1.26, 95% CI 0.95–1.67,

= 0.11). The associations of LND

with distant metastasis-free survival, cancer-specific sur-

vival (CSS), and overall survival among 370 PS matched

pairs are depicted in Supplementary

[41_TD$DIFF]

Figure 1

[16_TD$DIFF]

and

[17_TD$DIFF]

Fig. 1

As a sensitivity analysis to incorporate patients excluded

from PS analysis, we performed traditional Cox multivari-

able regression for the 1751 patients with nonmissing data.

After adjusting for all of the covariates used in PS

estimation, the HRs for the association of LND with distant

metastases, death from RCC, and death from any cause were

1.19 (95% CI 0.97–1.45,

= 0.10), 1.02 (95% CI 0.82–1.28,

= 0.85), and 1.09 (95% CI 0.93–1.28,

= 0.27), respectively.

We further evaluated whether LND was associated with

survival outcomes among patients at increased risk of

harboring pN1 disease. Among patients with preoperative

radiographic lymphadenopathy (cN1), LND was not signifi-

cantly associated with development of distant metastases

(adjusted for PS quintile: HR 0.71, 95% CI 0.33–1.54,

= 0.39; IPW: HR 0.75, 95% CI 0.38–1.46,

= 0.40), CSM

(adjusted for PS quintile: HR 0.94, 95% CI 0.41–2.15,

= 0.88; IPW: HR 0.94, 95% CI 0.46–1.92,

= 0.86), or

ACM (adjusted for PS quintile: HR 0.86, 95% CI 0.41–1.78,

= 0.68; IPW: HR 0.86, 95% CI 0.46–1.64,

= 0.65;

Table 3 )

Next, we used a logistic regression model (Supplementary

Table 3) to internally estimate the predicted probability of

pN1 disease for each patient in the cohort, regardless of LND

status, to evaluate the association of LND with oncologic

outcomes across increasing risk of pN1 disease. Among

patients with increasing threshold probabilities for pN1

disease ranging from 0.05 to 0.50, LND was not associated

with CSM or ACM

( Table 3

). Overall, LND was not associated

with the development of distant metastases, although two

thresholds demonstrated a modestly increased risk with

LND (0.05 probability with adjustment by PS quintile, and

0.15 probability with reweighting by IPWs).

In a sensitivity analysis, we examined the association of

extended LND (defined as

13 LNs removed) with

oncologic outcomes. Extended LND was not associated

Table 2 – Association of lymph node dissection with development of distant metastases, cancer-specific mortality (CSM), or all-cause

mortality (ACM) in the overall cohort (hazard ratio [HR] > 1 indicates an increased risk of event)

PS technique

Distant metastases

CSM

ACM

HR (95% CI)

value

HR (95% CI)

value

HR (95% CI)

value

Adjusted for PS quintile

1.25 (1.02–1.52)

0.03

1.15 (0.93–1.43)

0.21

1.08 (0.93–1.27)

0.32

Stratified by PS quintile

1.24 (1.03–1.51)

0.03

1.14 (0.92–1.41)

0.23

1.08 (0.92–1.26)

0.36

IPW

1.13 (0.95–1.36)

0.18

1.03 (0.84–1.26)

0.78

1.00 (0.87–1.14)

0.94

Matched pairs

1.26 (0.95–1.67)

0.11

1.25 (0.90–1.73)

0.19

1.09 (0.86–1.39)

0.46

CI = confidence interval; IPW = inverse probability weights; PS = propensity score.

E U R O P E A N U R O L O G Y 7 1 ( 2 0 1 7 ) 5 6 0 – 5 6 7

563