Abstract
Background:
Owing to surgical complexity and controversy regarding indications, there are wide practice variations in the use of postchemotherapy retroperitoneal lymph node dissection (PC-RPLND).
Objective:
To evaluate patterns of PC-RPLND use in the USA and evaluate the association between PC-RPLND and survival in advanced nonseminomatous germ cell tumors (NSGCTs).
Design, setting, and participants:
A retrospective, observational study using National Cancer Data Base (NCDB) data from 2004–2014 for 5062 men diagnosed with stage II/III NSGCT.
Outcome measurements and statistical analysis:
In a comparative analysis based on receipt of PC-RPLND, the primary outcome of interest was factors associated with omission of PC-RPLND as explored via logistic regression. As a secondary outcome, we evaluated the association between PC-RPLND and overall survival (OS) via multivariable Cox regression and propensity score matching (PSM).
Results and limitations:
Patients undergoing PC-RPLND were more likely to be younger, white, privately insured, and reside in more educated/wealthier regions (p < 0.001). Insurance status was independently associated with receipt of PC-RPLND; compared to patients with private insurance, those without insurance were significantly less likely to receive PC-RPLND (odds ratio 0.49; p < 0.001). After multivariate adjustment, age, comorbidity, non-private insurance, distance from hospital, clinical stage, and risk group were independently associated with all-cause mortality. In addition, omission of PC-RPLND remained associated with all-cause mortality (hazard ratio 1.98; p < 0.001). After PSM, the 5-yr OS was significantly lower among those not undergoing PC-RPLND (72% vs 77%; p = 0.007).
Conclusions:
PC-RPLND represents a critical part of the multidisciplinary management of NSGCT. Patients with non-private insurance are less likely to undergo PC-RPLND, and omission of PC-RPLND is associated with lower OS.
Patient summary:
We evaluated the practice patterns for advanced testicular cancer management and found that patients who did not undergo a postchemotherapy retroperitoneal lymph node dissection were more likely to have worse survival outcomes. Patients with unfavorable insurance were less likely to receive this surgical treatment.
Keywords: Testis cancer, Germ cell tumor, Surgery, Retroperitoneal lymph node dissection
1. Introduction
Primary metastatic or relapsed advanced nonseminomatous germ cell tumors (NSGCTs) are typically managed with upfront platinum-based chemotherapy followed by assessment of radiographic response to chemotherapy and measurement of serum tumor markers (STMs). For many patients, the next step in management is a postchemotherapy retroperitoneal lymph node dissection (PC-RPLND), which has the potential to be diagnostic of residual retroperitoneal mass, to treat residual disease, and is the only treatment for chemoresistant teratoma. For patients with residual masses >1 cm and normal STMs, the standard of care is PC-RPLND. Although advocated by many centers as a standard of care following chemotherapy, the indications for PC-RPLND are controversial in certain cases, in particular in cases with a complete response (CR), subcentimeter residual masses, or resistance to chemotherapy [1]. Furthermore, there is wide practice pattern variation in the use of PC-RPLND. We previously found that lower-volume hospitals are significantly less likely to perform PC-RPLND, and postulated that some of the variability may be because of concerns about the complexity of the operation and morbidity, as well as a lack of understanding regarding the biology of NSGCT [2].
Given the rarity of the disease, most publications on testicular germ cell tumor (TGCT) are single-center retrospective reviews, which are not necessarily reflective of nationwide practice patterns or socioeconomic factors that play a role in disease management. We used the National Cancer Data Base (NCDB) to study factors associated with the performance of PC-RPLND and the association between performance of PC-RPLND and survival in NSGCT.
2. Patients and methods
2.1. Data source
The NCDB is a national cancer registry sponsored by the American College of Surgeons (ACS) and the American Cancer Society that collects data on malignancies from ACS-Commission on Cancer (CoC) accredited facilities. It includes approximately 70% of all malignant cancers diagnosed in the USA from more than 1500 facilities [3]. The NCDB was queried for patients with TGCT diagnosed from 2004 to 2014.
2.2. Study population
There were 62 727 reported cases of testis cancer screened for inclusion. Figure 1 illustrates the case selection process. The International Classification of Disease for Oncology (3rd edition) was used to identify men diagnosed with NSGCT. Patients with non-testicular cancers, spermatocytic seminoma, sex cord/stromal tumors, seminoma, or unspecified germ cell tumors were excluded. Patients with unspecified American Joint Committee on Cancer (AJCC) clinical staging or stage 0–I disease were excluded. The CoC classifies patients according to the facility at which the malignancy was diagnosed and where the first-line treatment was provided. Patients who were not treated at the reporting facility (class of case “00”) were excluded. For patients treated at multiple CoC facilities, the NCDB reports the most recent treatment facility and/or the facility with the most complete records. Further information on CoC case classification and comparison of baseline demographics between patients included and those excluded because of missing or unknown selection parameters is provided in Supplementary Tables 1 and 2.
Fig. 1 –
Patient selection flow chart. NSGCT = nonseminomatous germ cell tumor; NOS = not otherwise specified; AJCC = American Joint Committee on Cancer; PC-RPLND = postchemotherapy retroperitoneal lymph node dissection.
2.3. Definition of PC-RPLND
Receipt of PC-RPLND is not explicitly listed in the NCDB, but it can be inferred. Patients with nonlocalized disease are typically managed with upfront chemotherapy. Patients given chemotherapy within 60 d of NSGCT diagnosis were categorized as having received primary chemotherapy. Within this group, patients recorded as having undergone regional lymph node surgery (RPLND) after chemotherapy were classified as having undergone PC-RPLND.
2.4. Covariates
Covariates included age, AJCC clinical stage, Charlson-Deyo comorbidity index, race/ethnicity, insurance coverage, distance from hospital, socioeconomic factors in the form of income and education in the patient’s region and a measure of urban versus rural geography [4], and International Germ Cell Cancer Collaborative Group (IGCCCG) risk classification [5]. The IGCCCG risk group was calculated by assessing metastatic location and postorchiectomy tumor marker status [5]. The CoC classifies hospitals according to facility type (eg, academic vs community) and geographic region; however, these variables are censored for patients younger than 39 yr. Given the young age at presentation of most men with testicular cancers, these variables were not included in the analysis.
2.5. Outcome
The primary outcome of interest was the association of PC-RPLND with overall survival (OS) measured from the time of diagnosis. As secondary outcomes, we analyzed patient and tumor factors associated with performance of PC-RPLND. In addition, we assessed the impact of PC-RPLND pathologic nodal staging on OS.
2.6. Statistical analysis
The mean ± standard deviation (SD) and median and interquartile range (IQR) are reported for normally and non-normally distributed continuous variables, respectively. Categorical and ordinal variables are presented as proportions. Baseline covariates were compared using the χ2 test for categorical variables and the Mann-Whitney U test for continuous and ordinal variables. A binomial logistic regression analysis was performed to ascertain the effects of covariates on the likelihood that PC-RPLND would be performed on a patient. Patients in the last year of the study (2014) were excluded from survival analysis, as directed by the NCDB. Survival was estimated using univariate analysis according to the Kaplan-Meier method, and comparisons were performed using the log-rank test and unadjusted Cox regression analysis. To minimize selection bias, we performed propensity score matching (PSM) to account for differences in covariates between patients who received PC-RPLND and those who did not. Matching was performed in a nearest 1:1 fashion according to the propensity to receive PC-RPLND according to multivariate logistic regression analysis of all observed covariates used in the multivariate analysis. The goal of this technique is to balance the covariates between those who did and did not receive PC-RPLND. Survival analysis was performed using Kaplan-Meier estimates of OS between the groups and univariate Cox regression analysis. Landmark analysis was performed at 6 mo after the time of diagnosis to account for the potential effect of immortal time bias, which may favor patients who went on to receive PC-RPLND.
Statistical analysis was performed using SPSS 22.0 (IBM, Armonk, NY, USA) and the MatchIt R package was used to perform PSM [6]. Boneferroni correction was applied for multiple comparison testing. Two-sided statistical significance was defined as p < 0.05.
3. Results
After applying the exclusion criteria, we identified 5062 patients for analysis. Table 1 describes the baseline patient characteristics. The mean ± SD patient age was 29.8 ± 10.1 yr. The majority of patients were non-Hispanic white (77.4%), and privately insured (62.2%). Clinical stage grouping was stage II for 1684 patients (32.9%) and stage III for 3439 patients (67.1%). The median follow-up from diagnosis was 41.6 mo (IQR 21.4–68.6). Of the overall cohort, 1230 (24.3%) underwent PC-RPLND and 3832 (75.7%) did not undergo PC-RPLND. Patients who underwent PC-RPLND were more likely to be younger (29.3 ± 9.5 vs 30.0 ± 10.3 yr; p = 0.015), white (81.6% vs 76.2%; p < 0.001), privately insured (72.6% vs 59.0%; p < 0.001), and to come from more highly educated/wealthier regions (p < 0.001). Patients who underwent PC-RPLND were also more likely to live further from their treating facility (median 17.8 miles, IQR 6.5–63.6 vs 9.7 miles, IQR 4.4–22.0; p < 0.001) and were less likely to have clinical stage III compared to stage II disease (61.5% vs 68.6%; p < 0.001).
Table 1 –
Patient characteristics for the unmatched cohort
| Characteristic | Overall | PC-RPLND |
|||
|---|---|---|---|---|---|
| No | Yes | p value | |||
| Patients (n) | 5062 | 3832 | 1230 | – | |
| Patient age (yr) | |||||
| Mean ± standard deviation | 29.8 ± 10.1 | 30.0 ± 10.3 | 29.3 ± 9.5 | 0.015 | |
| Median (IQR) | 28.0 (22.0–35.0) | 28.0 (23.0–35.0) | 27.0 (22.0–34.0) | 0.108 | |
| Charlson-Deyo comorbidity score, n (%) | |||||
| 0 | 4752 (92.8) | 3541 (92.4) | 1153 (93.7) | ||
| 1 | 323 (6.3) | 252 (6.6) | 68 (5.5) | 0.273 | |
| ≥2 | 48 (0.9) | 39 (1.0) | 9 (0.7) | ||
| Race/ethnicity, n (%) | |||||
| White (non-Hispanic) | 3929 (77.4) | 2894 (76.2) | 992 (81.6) | ||
| Black | 177 (3.5) | 148 (3.9) | 28 (2.3) | <0.001 | |
| Hispanic | 795 (15.7) | 625 (16.5) | 155 (12.8) | ||
| Other | 173 (3.4) | 131 (3.4) | 40 (3.3) | ||
| Insurance status, n (%) | |||||
| Private insurance | 3133 (62.2) | 2222 (59.0) | 879 (72.6) | ||
| Not insured | 138 (2.7) | 114 (3.0) | 22 (1.8) | ||
| Other government | 73 (1.4) | 58 (1.5) | 14 (1.2) | <0.001 | |
| Medicaid | 927 (18.4) | 737 (19.6) | 182 (15.0) | ||
| Medicare | 765 (15.2) | 635 (16.9) | 113 (9.3) | ||
| Zip-code education level, n (%) | |||||
| ≥21% NGHS | 934 (18.4) | 731 (19.3) | 184 (15.0) | ||
| 13–20.9% NGHS | 1200 (23.6) | 929 (24.5) | 262 (21.4) | <0.001 | |
| 7–12.9% NGHS | 1670 (32.9) | 1207 (31.9) | 446 (36.4) | ||
| <than 7% NGHS | 1270 (25.0) | 922 (24.3) | 332 (27.1) | ||
| Zip-code median household income, n (%) | |||||
| <$38 000 | 830 (16.4) | 655 (17.3) | 160 (13.1) | ||
| $38 000–$47 999 | 1208 (23.8) | 913 (24.1) | 277 (22.6) | <0.001 | |
| $48 000–$62 999 | 1420 (28.0) | 1070 (28.3) | 341 (27.9) | ||
| ≥$63 000 | 1613 (31.8) | 1148 (30.3) | 446 (36.4) | ||
| Distance from hospital (miles) | |||||
| Mean ± standard deviation | 41.6 ± 144.6 | 26.9 ± 101.1 | 87.0 ± 227.5 | <0.001 | |
| Median (IQR) | 10.9 (4.7–27.2) | 9.7 (4.4–22.0) | 17.8 (6.5–63.6) | <0.001 | |
| Household location, n (%) | |||||
| Metropolitan | 4240 (85.5) | 3154 (85.1) | 1032 (86.6) | ||
| Urban | 639 (12.9) | 497 (13.4) | 138 (11.6) | 0.190 | |
| Rural | 79 (1.6) | 55 (1.5) | 22 (1.8) | ||
| AJCC clinical nodal stage, n (%) | |||||
| cN0 | 708 (14.0) | 620 (16.3) | 76 (6.2) | ||
| cN1–3 | 3856 (76.0) | 2759 (72.7) | 1066 (87.3) | ||
| cN1 | 1368 (27.0) | 1087 (28.7) | 269 (22.0) | <0.001 | |
| cN2 | 1410 (27.8) | 1012 (26.7) | 390 (31.9) | ||
| cN3 | 1078 (21.2) | 660 (17.4) | 407 (33.3) | ||
| cNX | 509 (10.0) | 414 (10.9) | 79 (6.5) | ||
| AJCC clinical metastasis stage, n (%) | |||||
| cM0 | 2906 (57.6) | 2131 (56.5) | 751 (61.9) | ||
| cM1 | 2140 (42.4) | 1643 (43.5) | 463 (38.1) | ||
| cM1 (NOS) | 745 (14.8) | 589 (15.6) | 141 (11.6) | 0.001 | |
| cM1a | 877 (17.4) | 653 (17.3) | 211 (17.4) | ||
| cM1b | 518 (10.3) | 401 (10.6) | 111 (9.1) | ||
| AJCC clinical stage, n (%) | |||||
| Stage II | 1684 (32.9) | 1203 (31.4) | 474 (38.5) | <0.001 | |
| Stage III | 3439 (67.1) | 2629 (68.6) | 756 (61.5) | ||
| IGCCCG risk group, n (%) | |||||
| Good risk | 779 (15.2) | 547 (14.3) | 231 (18.8) | <0.001 | |
| Intermediate risk | 232 (4.5) | 152 (4.0) | 76 (6.2) | ||
| Poor risk | 677 (13.2) | 523 (13.6) | 146 (11.9) | ||
| Unable to determine | 3435 (67.1) | 2610 (68.1) | 777 (63.2) | ||
SD = standard deviation; IQR = interquartile range; NGHS = not graduating high school; AJCC = American Joint Committee on Cancer; IGCCCG = International Germ Cell Collaborative Group.
On multivariate logistic regression, patients with clinical stage III disease were less likely to receive PC-RPLND (odds ratio [OR] 0.74, 95% confidence interval [CI] 0.63–0.88; p = 0.001). Compared to patients with good risk, those with intermediate-risk disease were more likely to undergo PC-RPLND (OR 1.74, 95% CI 1.17–2.59; p = 0.006). In addition, distance from a patient’s residence to the hospital location was significantly associated with receiving PC-RPLND (OR 1.55, 95% CI 1.46–1.65; p < 0.001). Patient insurance status was independently associated with receipt of PC-RPLND; compared to patients with private insurance, those with Medicaid (OR 0.72, 95% CI 0.58–0.90; p = 0.003) or no insurance (OR 0.49, 95% CI 0.38–0.63; p < 0.001) were significantly less likely to receive PC-RPLND. After adjusting for other variables, patient race/ethnicity, comorbidity score, age, and zip-code income and education level were not associated with receipt of PC-RPLND (Table 2). All-cause mortality was reported for 725 patients. Table 3 lists results for univariate and multivariate survival analyses. On unadjusted univariate Cox regression analysis, patient age, higher comorbidity, minority race/ethnicity, non-private insurance coverage, lower median household income and education level, distance from hospital, clinical stage III (vs stage II), IGCCCG risk group other than good risk, and omission of PC-RPLND were significantly associated with higher risk of all-cause mortality. The 5-yr OS was 90% for those who underwent PC-RPLND versus 79% for those who did not (p < 0.001; Fig. 2A). After multivariate adjustment; age, comorbidity status, non-private insurance, distance from hospital, clinical stage, and IGCCCG risk group remained independently associated with all-cause mortality. In addition, omission of PC-RPLND remained significantly associated with all-cause mortality with (hazard ratio [HR] 1.98, 95% CI 1.56–2.51; p < 0.001).
Table 2 –
Factors associated with receipt of postchemotherapy reptroperitoneal lymph node dissection
| Characteristic | OR (95% CI) | p value |
|---|---|---|
| Patient age in years a | 0.95 (0.74–1.20) | 0.649 |
| Charslon-Deyo comorbidity score | ||
| 0 | Reference | – |
| 1 | 0.78 (0.56–1.09) | 0.149 |
| ≥2 | 0.98 (0.43–2.23) | 0.967 |
| Race/ethnicity | ||
| White (non-Hispanic) | Reference | – |
| Black | 0.73 (0.44–1.20) | 0.215 |
| Hispanic | 0.96 (0.75–1.23) | 0.759 |
| Other | 0.90 (0.58–1.41) | 0.646 |
| Insurance status | ||
| Private | Reference | – |
| Medicare | 0.70 (0.41–1.19) | 0.192 |
| Other government | 0.44 (0.21–0.94) | 0.034 |
| Medicaid | 0.72 (0.58–0.90) | 0.003 |
| Not insured | 0.49 (0.38–0.63) | <0.001 |
| Zip-code median household income | ||
| <$38 000 | Reference | – |
| $38 000–$47 999 | 1.02 (0.78–1.33) | 0.883 |
| $48 000–$62 999 | 0.97 (0.73–1.28) | 0.811 |
| ≥$63 000 | 1.16 (0.85–1.59) | 0.342 |
| Zip-code education level | ||
| ≥21% not graduating high school | Reference | – |
| 13–20.9% not graduating high school | 0.90 (0.69–1.18) | 0.446 |
| 7–12.9% not graduating high school | 1.09 (0.83–1.43) | 0.523 |
| <7% not graduating high school | 0.94 (0.69–1.30) | 0.720 |
| Distance from hospital in miles a | 1.55 (1.46–1.65) | <0.001 |
| Household location | ||
| Metropolitan | Reference | – |
| Urban | 0.53 (0.41–0.68) | <0.001 |
| Rural | 0.60 (0.33–1.07) | 0.083 |
| American Joint Committee on Cancer clinical stage | ||
| Stage II | Reference | – |
| Stage III | 0.74 (0.63–0.88) | 0.001 |
| IGCCCG risk group | ||
| Good risk | Reference | – |
| Intermediate risk | 1.74 (1.17–2.59) | 0.006 |
| Poor risk | 0.75 (0.55–1.03) | 0.076 |
| Unable to determine | 0.87 (0.70–1.07) | 0.192 |
OR = odds ratio; CI = confidence interval; IGCCCG = International Germ Cell Collaborative Group.
Natural log–transformed.
Table 3 –
Factors associated with all-cause mortality
| Characteristic | Univariate analysis |
Multivariate analysis |
||
|---|---|---|---|---|
| HR (95% CI) | p value | HR (95% CI) | p value | |
| Patient age (years) | 1.034 (1.028–1.041) | <0.001 | 1.03 (1.02–1.04) | <0.001 |
| Charlson-Deyo comorbidity score | ||||
| 0 | Reference | – | Reference | – |
| 1 | 1.977 (1.551–2.519) | <0.001 | 1.43 (1.10–1.85) | 0.007 |
| ≥2 | 3.643 (2.250–5.898) | <0.001 | 3.25 (1.98–5.32) | <0.001 |
| Race/ethnicity | ||||
| White (non-Hispanic) | Reference | – | Reference | – |
| Black | 2.247 (1.657–3.047) | <0.001 | 1.37 (0.97–1.92) | 0.072 |
| Hispanic | 1.325 (1.088–1.614) | 0.005 | 0.97 (0.77–1.23) | 0.821 |
| Other | 1.732 (1.224–2.450) | 0.002 | 1.56 (1.08–2.25) | 0.017 |
| Insurance status | ||||
| Private insurance | Reference | – | Reference | – |
| Medicaid | 2.650 (1.842–3.812) | <0.001 | 1.52 (1.02–2.26) | 0.037 |
| Medicare | 2.218 (1.299–3.790) | 0.004 | 1.53 (0.87–2.70) | 0.141 |
| Other government | 2.357 (1.971–2.818) | <0.001 | 1.86 (1.53–2.26) | <0.001 |
| Not insured | 2.151 (1.772–2.611) | <0.001 | 1.74 (1.41–2.15) | <0.001 |
| Zip-code median household income | ||||
| <$38 000 | Reference | – | Reference | – |
| $38 000–$47 999 | 0.784 (0.631–0.972) | 0.027 | 1.05 (0.82–1.33) | 0.719 |
| $48 000–$62 999 | 0.699 (0.565–0.864) | 0.001 | 1.10 (0.85–1.43) | 0.470 |
| ≥$63 000 | 0.564 (0.454–0.700) | <0.001 | 0.96 (0.71–1.31) | 0.803 |
| Zip-code education level | ||||
| ≥21% NGHS | Reference | – | Reference | – |
| 13–20.9% NGHS | 0.81 (0.66–1.00) | 0.048 | 1.04 (0.82–1.31) | 0.774 |
| 7–12.9% NGHS | 0.65 (0.53–0.79) | <0.001 | 0.88 (0.68–1.15) | 0.349 |
| <7% NGHS | 0.54 (0.54–0.68) | <0.001 | 0.82 (0.60–1.13) | 0.233 |
| Distance from hospital in miles | 1.000 (1.000–1.001) | 0.042 | 1.00 (1.00–1.00) | 0.001 |
| Household location | ||||
| Metropolitan | Reference | – | Reference | – |
| Urban | 1.15 (0.93–1.42) | 0.181 | 1.08 (0.86–1.36) | 0.513 |
| Rural | 1.25 (0.72–2.17) | 0.441 | 0.76 (0.40–1.44) | 0.400 |
| AJCC clinical stage | ||||
| Stage II | Reference | – | Reference | – |
| Stage III | 6.312 (4.836–8.238) | <0.001 | 4.17 (3.13–5.54) | <0.001 |
| IGCCCG risk group | ||||
| Good risk | Reference | – | Reference | – |
| Intermediate risk | 2.92 (1.55–5.50) | 0.001 | 1.51 (0.77–2.96) | 0.229 |
| Poor risk | 11.69 (7.51–18.21) | <0.001 | 5.51 (3.51–8.66) | 0.000 |
| Unknown | 3.97 (2.59–6.08) | <0.001 | 2.61 (1.69–4.01) | 0.000 |
| Performance of PC-RPLND | ||||
| Yes | Reference | – | Reference | – |
| No | 2.492 (1.985–3.130) | <0.001 | 1.98 (1.56–2.51) | <0.001 |
HR = hazard ratio; CI = confidence interval; NGHS = not graduating high school; AJCC = American Joint Committee on Cancer; IGCCCG = International Germ Cell Collaborative Group, PC-RPLND = post-chemotherapy retroperitoneal lymph node dissection
Fig. 2 –
Overall survival (OS) stratified by receipt of postchemotherapy retroperitoneal lymph node dissection (PC-RPLND) for (A) the unmatched cohort; (B) the propensity score–matched cohort; and (C) the propensity score–matched cohort with 6-mo landmark analysis.
Of the 1230 patients who underwent PC-RPLND, pathologic nodal stage was not recorded for 360 patients (29.3%), with pN0 in 300 patients (24.4%), pN1 in 153 patients (12.4%), pN2 in 192 patients (15.6%), and pN3 in 215 patients (17.5%). Among patients with available pathologic data, 65.1% had positive nodes on PC-RPLND. When assessing OS by pathologic nodal stage, those who did not undergo PC-RPLND had significantly worse OS (HR for all-cause mortality 1.68, 95% CI 1.30–2.16; p < 0.001) compared to pN0 patients (Fig. 3). All-cause mortality did not differ significantly between patients with pathologically positive nodes (pN1–3) and those with negative nodes (pN0) following PC-RPLND (HR 1.05, 95% CI 0.79–1.41; p = 0.732).
Fig. 3 –
Overall survival (OS) stratified by pathologic nodal status. PC-RPLND = postchemotherapy retroperitoneal lymph node dissection.
After PSM, there were 2278 patients in the analysis evenly split between those who underwent PC-RPLND (n = 1139) and those who did not (n = 1139). Table 4 demonstrates that the baseline characteristics between the groups after PSM were evenly distributed, with the exception of distance from hospital. The 5-yr OS was 77% for those who underwent PC-RPLND versus 72% for those who did not (p = 0.007; Fig. 2B).
Table 4 –
Patient characteristics for the propensity score–matched cohort
| Characteristic | Overall | PC-RPLND |
||
|---|---|---|---|---|
| No | Yes | p value | ||
| Patients (n) | 2278 | 1139 | 1139 | – |
| Mean age ± SD (yr) | 29.3 ± 9.6 | 29.3 ± 9.7 | 29.3 ± 9.6 | 0.73 |
| Charlson-Deyo comorbidity score | ||||
| 0 | 2133 (93.6) | 1067 (93.7) | 1066 (93.6) | 0.958 |
| 1 | 126 (5.5) | 62 (5.4) | 64 (5.6) | |
| ≥2 | 19 (0.8) | 10 (0.9) | 9 (0.8) | |
| Race/ethnicity | ||||
| White (non-Hispanic) | 2133 (93.6) | 928 (81.5) | 935 (82.1) | 0.908 |
| Black | 126 (5.5) | 22 (1.9) | 25 (2.2) | |
| Hispanic | 19 (0.8) | 156 (13.7) | 146 (12.8) | |
| Other | 2133 (93.6) | 33 (2.9) | 33 (2.9) | |
| Insurance status, n (%) | ||||
| Private insurance | 1616 (70.9) | 782 (68.7) | 834 (73.2) | 0.188c |
| Not insured | 44 (1.9) | 24 (2.1) | 20 (1.8) | |
| Other government | 25 (1.1) | 12 (1.1) | 13 (1.1) | |
| Medicaid | 366 (16.1) | 199 (17.5) | 167 (14.7) | |
| Medicare | 227 (10.0) | 122 (10.7) | 105 (9.2) | |
| Zip-code education level, n (%) | ||||
| ≥21% NGHS | 359 (15.8) | 187 (16.4) | 172 (15.1) | 0.786 |
| 13–20.9% NGHS | 468 (20.5) | 227 (19.9) | 241 (21.2) | |
| 7–12.9% NGHS | 838 (36.8) | 420 (36.9) | 418 (36.7) | |
| <7% NGHS | 613 (26.9) | 305 (26.8) | 308 (27.0) | |
| Zip-code median household income, n (%) | ||||
| <$38 000 | 304 (13.3) | 155 (13.6) | 149 (13.1) | 0.577 |
| $38 000–$47 999 | 532 (23.4) | 273 (24.0) | 259 (22.7) | |
| $48 000–$62 999 | 603 (26.5) | 287 (25.2) | 316 (27.7) | |
| ≥$63000 | 839 (36.8) | 424 (37.2) | 415 (36.4) | |
| Mean distance from hospital ± SD (miles) | 2.9 ± 1.5 | 2.8 ± 1.3 | 3.0 ± 1.6 | <0.001 |
| Household location, n (%) | ||||
| Metropolitan | 1967 (86.3) | 978 (85.9) | 989 (86.8) | 0.795 |
| Urban | 268 (11.8) | 139 (12.2) | 129 (11.3) | |
| Rural | 43 (1.9) | 22 (1.9) | 21 (1.8) | |
| AJCC clinical nodal stage, n (%) | ||||
| cN0 | 153 (6.7) | 81 (7.1) | 72 (6.3) | 0.479 |
| cN1 | 519 (22.8) | 271 (23.8) | 248 (21.8) | |
| cN2 | 745 (32.7) | 375 (32.9) | 370 (32.5) | |
| cN3 | 723 (31.7) | 349 (30.6) | 374 (32.8) | |
| cNX | 138 (6.1) | 63 (5.5) | 75 (6.6) | |
| AJCC clinical metastasis stage, n (%) | ||||
| cM0 | 1413 (62.0) | 701 (61.5) | 712 (62.5) | 0.814 |
| cM1 | 865 (38.0) | |||
| cM1 (NOS) | 270 (11.9) | 142 (12.5) | 128 (11.2) | – |
| cM1a | 386 (16.9) | 190 (16.7) | 196 (17.2) | – |
| cM1b | 209 (9.2) | 106 (9.3) | 103 (9.0) | – |
| AJCC clinical stage | ||||
| Stage II | 849 (37.3) | 411 (36.1) | 438 (38.5) | 0.242 |
| Stage III | 1429 (62.7) | 728 (63.9) | 701 (61.5) | |
| IGCCCG risk group, n (%) | ||||
| Good risk | 410 (18.0) | 195 (17.1) | 215 (18.9) | 0.719 |
| Intermediate risk | 149 (6.5) | 73 (6.4) | 76 (6.7) | |
| Poor risk | 266 (11.7) | 135 (11.9) | 131 (11.5) | |
| Unknown | 1453 (63.8) | 736 (64.6) | 717 (62.9) | |
SD = standard deviation; NGHS = not graduating high school; AJCC = American Joint Committee on Cancer; IGCCCG = International Germ Cell Collaborative Group; PC-RPLND = postchemotherapy retroperitoneal lymph node dissection.
In a landmark analysis performed at 6 mo after NSGCT diagnosis in the PSM cohort, the survival difference between PC-RPLND and no PC-RPLND remained unchanged relative to the “naïve” analysis. A total of 398 patients (17.5%) were excluded from the landmark analysis because of <6 mo of follow-up due to death or censoring, leaving 1880 patients for analysis (n = 935 for no PC-RPLND, n = 945 for PC-RPLND). The 5-yr OS was 88% for the PC-RPLND group versus 82% for the no PC-RPLND group (p < 0.001; Fig. 2C).
4. Discussion
PC-RPLND plays a critical role in the multidisciplinary approach to patients with advanced testicular cancer. The rationale for PC-RPLND is to remove persistent lymph nodes that may contain mature teratoma in 30–40% or viable tumor in 10–20% of patients [7–10]. PC-RPLND not only provides crucial diagnostic information on the residual mass but can also have therapeutic benefit, and is the only definitive way of treating teratoma [11,12]. Approximately 70% of patients who present with metastatic NSGCT to the retroperitoneum achieve CR to chemotherapy, defined as normalization of STMs and radiographic resolution of the retroperitoneal mass [13]. Regarding the remaining 30% of patients with persistent radiographic masses and negative STMs, there is universal agreement that PC-RPLND is the treatment of choice because these masses can harbor viable GCT or teratoma. However, there is considerable debate regarding the definition of a persistent radiographic mass and the management of patients who achieve CR. Several groups reported that even with complete radiographic response to chemotherapy, 20–30% of patients had residual teratoma and 1–9% had viable GCT, demonstrating that the retroperitoneum remains difficult to accurately stage in the postchemotherapy setting [10,14–16]. Thus, several groups have attempted to develop preoperative clinical tools to predict the histology of postchemotherapy residual masses; however, these attempts have had limited success because of high false-negative rates [8,17–20]. Given the unpredictable behavior of subcentimeter residual masses, some centers recommend PC-RPLND even for patients who achieve CR [10,16]. Despite guidelines to standardize the treatment of NSGCT, institutional treatment variations continue to occur [2]. Our study sought to analyze factors associated with the performance of PC-RPLND and the association between PC-RPLND and survival. After multivariable adjustment, lack of private insurance remained significantly associated with omission of PC-RPLND. This is troubling, as it is unlikely that patients without insurance have any biologic differences in chemoresponsiveness to explain differences in PC-RPLND use. In addition, we found that patients with clinical stage III and IGCCCG poor risk were less likely to receive PC-RPLND. We speculate this may be because these patients have higher rates of platinum-resistant disease, necessitating salvage chemotherapy rather than PC-RPLND. Patients who underwent PC-RPLND were more likely to travel further to seek care, supporting centralization of TGCT management.
A concerning finding is that patients who did not undergo PC-RPLND had a twofold higher risk of all-cause mortality (HR 1.98; p < 0.001). This is consistent with a report by Fosså et al [21], who described significantly higher mortality among NSGCT patients who did not undergo RPLND. Unfortunately, the NCDB does not allow determination of response to chemotherapy; therefore, no conclusions can be made regarding use of PC-RPLND among patients who achieved CR to chemotherapy, nor were we able to substratify patients according to residual mass size. Nevertheless, even among patients with CR observed by Ehrlich et al [22], the relapse rate in the retroperitoneum was 4%, and ultimately one-third of those who experienced relapse died of their disease. To take this one step further, if we assume that a significant proportion of patients had CR and PC-RPLND was intentionally omitted after shared decision-making that considered the risks and benefits of PC-RPLND, we still observed a dramatic effect on OS in the subgroup/minority of patients with residual disease who would have most likely benefited from PC-RPLND.
Recent data show that 17% of the young population in the USA remain uninsured [23]. Consistent with previous reports on socioeconomic disparities in testicular cancer [21,24,25], we find that unfavorable insurance status is an important determinant of survival. While some of this effect may be explained by delayed presentation and patient willingness to accept intensive treatment or monitoring, it is likely that there are also differences in how physicians approach management for insured versus uninsured patients. It is an unfortunate reality in our society that uninsured patients are more likely to present with metastatic disease, be undertreated, and die after a diagnosis of cancer [26]. These are especially alarming findings with respect to testicular cancer given its success story and potential for high cure rates. Ultimately, reductions in insurance gaps are needed to improve TGCT outcomes.
There are several limitations to the study that should be acknowledged. The NCDB does not provide information on response to chemotherapy, recurrence, or RPLND histology. In addition, performance of PC-RPLND is not specifically coded, but can rather be inferred from performance of regional lymph node surgery following chemotherapy. As mentioned previously, facility location and type were not analyzed as variables in our analysis as these are censored for patients younger than 39 yr because of privacy concerns. The NCDB does not allow for assessment of cancer-specific mortality; however, in TGCT this is less of a concern as the vast majority of patients are young and healthy, with few competing causes of mortality.
5. Conclusions
Our results suggest that unfavorable insurance status is independently associated with omission of PC-RPLND. In addition, we found that patients who do not undergo PC-RPLND are have a higher risk of mortality.
Supplementary Material
Postchemotherapy retroperitoneal lymph node dissection (PC-RPLND) is an integral part of nonseminomatous germ cell tumor management. We found that patients with unfavorable insurance were less likely to receive PC-RPLND and omission of this surgical treatment was independently associated with worse survival.
Acknowledgments
Financial disclosures: Aditya Bagrodia certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.
Funding/Support and role of the sponsor: This work was supported by the National Institute of Health (T32 CA136515 Ruth L. Kirschstein Institutional National Research Award to S.L.W.) and the Dedman Family Scholarship in Clinical Care (to A.B.). The sponsors played no direct role in the study.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Gilligan T, Beard C, Carnerio B, et al. NCCN clinical practice guidelines in oncology: testicular cancer Fort Washington, PA: National Comprehensive Cancer Network; 2017. [DOI] [PubMed] [Google Scholar]
- 2.Woldu SL, Matulay JT, Clinton TN, et al. Impact of hospital case volume on testicular cancer outcomes and practice patterns. Urol Oncol 2018;36:14. [DOI] [PubMed] [Google Scholar]
- 3.Bilimoria KY, Stewart AK, Winchester DP, Ko CY. The National Cancer Data Base: a powerful initiative to improve cancer care in the United States. Ann Surg Oncol 2008;15:683–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.US Department of Agriculture Economic Research Service. Rural-urban continuum codes 2013 www.ers.usda.gov/data-products/rural-urban-continuum-codes
- 5.International Germ Cell Cancer Collaborative Group. International Germ Cell Consensus Classification: a prognostic factor-based staging system for metastatic germ cell cancers. J Clin Oncol 1997;15:594–603. [DOI] [PubMed] [Google Scholar]
- 6.Ho DE, Imai K, King G, Stuart EA. MatchIt: nonparametric preprocessing for parametric causal inference. J Stat Softw 2011;42:1–28. [Google Scholar]
- 7.Krege S, Beyer J, Souchon R, et al. European consensus conference on diagnosis and treatment of germ cell cancer: a report of the second meeting of the European Germ Cell Cancer Consensus Group (EGCCCG): part II. Eur Urol 2008;53:497–513. [DOI] [PubMed] [Google Scholar]
- 8.Heidenreich A, Pfister D. Retroperitoneal lymphadenectomy and resection for testicular cancer: an update on best practice. Ther Adv Urol 2012;4:187–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Albers P, Weissbach L, Krege S, et al. Prediction of necrosis after chemotherapy of advanced germ cell tumors: results of a prospective multicenter trial of the German Testicular Cancer Study Group. J Urol 2004;171:1835–8. [DOI] [PubMed] [Google Scholar]
- 10.Oldenburg J, Alfsen GC, Lien HH, Aass N, Waehre H, Fossa SD. Postchemotherapy retroperitoneal surgery remains necessary in patients with nonseminomatous testicular cancer and minimal residual tumor masses. J Clin Oncol 2003;21:3310–7. [DOI] [PubMed] [Google Scholar]
- 11.Ahmed T, Bosl GJ, Hajdu SI. Teratoma with malignant transformation in germ cell tumors in men. Cancer 1985;56:860–3. [DOI] [PubMed] [Google Scholar]
- 12.Fizazi K, Tjulandin S, Salvioni R, et al. Viable malignant cells after primary chemotherapy for disseminated nonseminomatous germ cell tumors: prognostic factors and role of postsurgery chemotherapy—results from an international study group. J Clin Oncol 2001;19:2647–57. [DOI] [PubMed] [Google Scholar]
- 13.Donohue JP, Leviovitch I, Foster RS, Baniel J, Tognoni P. Integration of surgery and systemic therapy: results and principles of integration. Semin Urol Oncol 1998;16:65–71. [PubMed] [Google Scholar]
- 14.Fossa SD, Ous S, Lien HH, Stenwig AE. Post-chemotherapy lymph node histology in radiologically normal patients with metastatic nonseminomatous testicular cancer. J Urol 1989;141:557–9. [DOI] [PubMed] [Google Scholar]
- 15.Carver BS, Bianco FJ Jr, Shayegan B, et al. Predicting teratoma in the retroperitoneum in men undergoing post-chemotherapy retroperitoneal lymph node dissection. J Urol 2006;176:100–3. [DOI] [PubMed] [Google Scholar]
- 16.Carver BS, Shayegan B, Eggener S, et al. Incidence of metastatic nonseminomatous germ cell tumor outside the boundaries of a modified postchemotherapy retroperitoneal lymph node dissection. J Clin Oncol 2007;25:4365–9. [DOI] [PubMed] [Google Scholar]
- 17.Beck SD, Foster RS, Bihrle R, et al. Teratoma in the orchiectomy specimen and volume of metastasis are predictors of retroperitoneal teratoma in post-chemotherapy nonseminomatous testis cancer. J Urol 2002;168:1402–4. [DOI] [PubMed] [Google Scholar]
- 18.Steyerberg EW, Keizer HJ, Fossa SD, et al. Prediction of residual retroperitoneal mass histology after chemotherapy for metastatic nonseminomatous germ cell tumor: multivariate analysis of individual patient data from six study groups. J Clin Oncol 1995;13:1177–87. [DOI] [PubMed] [Google Scholar]
- 19.Steyerberg EW, Gerl A, Fossa SD, et al. Validity of predictions of residual retroperitoneal mass histology in nonseminomatous testicular cancer. J Clin Oncol 1998;16:269–74. [DOI] [PubMed] [Google Scholar]
- 20.Vergouwe Y, Steyerberg EW, Foster RS, Habbema JD, Donohue JP. Validation of a prediction model and its predictors for the histology of residual masses in nonseminomatous testicular cancer. J Urol 2001;165:84–8. [DOI] [PubMed] [Google Scholar]
- 21.Fosså SD, Cvancarova M, Chen L, et al. Adverse prognostic factors for testicular cancer– specific survival: a population-based study of 27,948 patients. J Clin Oncol 2011;29:963– 70. [DOI] [PubMed] [Google Scholar]
- 22.Ehrlich Y, Brames MJ, Beck SD, Foster RS, Einhorn LH. Long-term follow-up of cisplatin combination chemotherapy in patients with disseminated nonseminomatous germ cell tumors: is a postchemotherapy retroperitoneal lymph node dissection needed after complete remission? J Clin Oncol 2010;28:531–6. [DOI] [PubMed] [Google Scholar]
- 23.Cohen RA, Martinez ME, Zammitti EP. Health insurance coverage: early release of estimates from the National Health Interview Survey, January–March 2017. Hyattsville, MD: National Center for Health Statistics; 2017. [Google Scholar]
- 24.Sun M, Abdollah F, Liberman D, et al. Racial disparities and socioeconomic status in men diagnosed with testicular germ cell tumors: a survival analysis. Cancer 2011;117:4277–85. [DOI] [PubMed] [Google Scholar]
- 25.Kamel MH, Elfaramawi M, Jadhav S, Saafan A, Raheem OA, Davis R. Insurance status and differences in treatment and survival of testicular cancer patients. Urology 2016;87:140–5. [DOI] [PubMed] [Google Scholar]
- 26.Aizer AA, Falit B, Mendu ML, et al. Cancer-specific outcomes among young adults without health insurance. J Clin Oncol 2014;32:2025–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.